RESEARCH
GRANTS
STUDIES ON ICE FOG
U. S. ENVIRONMENTAL PROTECTION AGENCY
-------
The APTD (Air Pollution Technical Data) series of reports is issued
to report technical data of interest to a limited readership. Copies of
APTD reports are available free of charge - as supplies permit - from
the Office of Technical Information and Publications, Environmental
Protection Agency, Research Triangle Park, North Carolina 27711.
This report was furnished to the Environmental Protection Agency by
the Geophysical Institute of the University of Alaska in fulfillment of
Research Grant No. AP-00449. The contents of the report are repro-
duced herein as-received from the contractor. The opinions, findings,
and conclusions expressed are those of the author and not necessarily
those of the Environmental Protection Agency.
Office of Air Programs Publication No. APTD-0626
-------
GEOPtCYSICAL INSTITUTE
of the
UNIVERSITY OF ALASKA
STUDIES ON ICE FOG
by
Takeshi Ohtake
Final Report
AP-00449
June 1970
This document has been approved
for public release and sale; its
distribution is unlimited.
Prepared for
National Center for Air Pollution Control
Public Health Service
Department of Health,, Education and Welfare
Principal Investigator:
Takeshi Ohtake
Approved by:
i -'' O . / -/
N- ..*». • L (
Keith B. Mather
Director
-------
STUDIES ON ICE FOG
Takeshi Ohtake
Geophysical Institute
of the
University of Alaska
ENVIRONMENTAL PROTECTION AGENCY
Research Triangle Park, North Carolina
July 1971
-------
ABSTRACT
In order to clarify the mechanism of ice-fog formation, various
atmospheric factors in ice fogs such as size and concentration of i:;e-
fog crystals, condensation nuclei and ice nuclei, amount of water vapor,
temperature profile near the sources of ice fog, etc. were measured.
Nuclei of the ice-fog crystals were studied by use of an electron
microscope and electron-diffraction. The examination showed that most
nuclei of ice-fog crystals were combustion by-products and many indi-
vidual crystals collected near open water did not have a nucleus,
especially at temperatures below -40C. Dust particles or particles
from air pollution are not essential for formation of ice fog; they merely
stimulate freezing of water droplets at higher temperatures than the
spontaneous freezing temperature. The essential factor is to first form
many water droplets in the atmosphere through condensation of water vapor.
Based on these measurements and calculations of time required for
water droplets to freeze, a physical mechanism of ice fog formation is
proposed as follows: 1) Water vapor coming from open water which is
exposed to a low temperature atmosphere, plus water vapor from various
exhausts of combustion processes is released into the almost ice-saturated
atmosphere and condenses into water droplets, 2) The droplets freeze
very shortly after their formation and before entirely evaporating,
3) Such ice particles do not evaporate or grow much and stay in the
atmosphere with insignificant fall out, and 4) These processes operate
more efficiently in colder environments, which make ice fog more serious
at lower temperatures.
-------
Measurements of humidity and evaporation rates from open water at
low air temperatures are described, as are some phenomena relating to
ice crystals found in ice fog. The types of meteorological situations
which are litcely to cause ice fog, visual range in ice fog, and the
electric properties and theoretical studies of size distribution of ice
fog crystals are also described.
ii
-------
TABLE OF CONTENTS
Page
ABSTRACT i
PREFACE AND ACKNOWLEDGEMENTS vii
LIST OF PUBLICATIONS ix
LIST OF TABLES x
LIST OF FIGURES xi
1. INTRODUCTION 1
2. MEASUREMENTS OF ICE-FOG CRYSTALS 7
3. MEASUREMENTS OF CONDENSATION NUCLEI 20
4. MEASUREMENT OF ICE NUCLEI 25
5. RELATIONSHIP BETWEEN CONDENSATION AND ICE NUCLEI AND 32
ICE-FOG CRYSTALS IN THEIR CONCENTRATIONS
6. ELECTRON MICROSCOPE STUDIES OF STEAM-FOG AND ICE-FOG 34
CRYSTALS AND THEIR NUCLEI
a. Smoke Samples 35
b. Steam-Fog Nuclei 37
c. Ice-Fog Crystals and Their Nuclei 37
1. Method of Making Specimens 37
2. Size and Shape of Ice-Fog Crystals 41
3. Sizes and Composition of Ice-Fog Nuclei and 41
Electric Conductivity of Melted Ice-Fog Crystals
i
4. Position of Nuclei in Ice-Fog Crystals 50
5. Spicules on Ice Fog Crystals 53
6. Sintering of Ice-Fog Crystals and Structure of 55
Ice-Fog Crystals
d. Inactive Nuclei in Ice Fog 61
iii
-------
TABLE OF CONTENTS
(Cont'd)
Page
7. OPTICAL MICROSCOPE STUDIES OF ICE-FOG CRYSTALS 62
A. THE EFFECT OF TEMPERATURE AND HUMIDITY ON ICE-FOG 62
CRYSTALS
1. Sampling Method for Ice-Fog Crystals 62
2. Method of Analysis 63
a. Determination of the Diameters 63
b. Classification of the Shapes of Ice-Fog 63
Crystals
3. Results and Discussion 64
a. Mean Diameters of Ice-Fog Crystals 64
b. Size Distribution and Percentage Distribution 72
of Types of Ice-Fog Crystals
c. Numbers of Precipitated Ice-Fog Crystals 72
d. Precipitation Rate and Solid Water Content 78
of Ice-Fog Crystals
e. The Effect of Ambient Humidity on Ice-Fog 83
Crystals
f. Ice-Fog Crystals from an Unpolluted Area with 85
a Moisture Source
B. UNUSUAL .CRYSTALS IN ICE FOG (POLYHEDRAL ICE CRYSTALS) 87
8. MECHANISM OF ICE FOG FORMATION 97
a. Measurements of Humidity under Low Temperature 98
.Conditions
i. Data from Hair Hygrometers 99
i
ii. Absolute Method of Humidity Measurement 102
b. Visual Observations of Ice-Fog Sources 109
lv
-------
TABLE OF CONSENTS
(Cont'd)
Page
c. Measurement of Evaporation of Water from the River or 117
a Pan and the Formation of Water Droplets from the Vapor
d. Temperature Measurement above Water Surface 120
e. Derived Supersaturation Degrees, Compared with Super- 122
saturations Required to Activate Condensation Nuclei
f. Tune Required for Droplets to Freeze 123
1. Conductive Cooling 123
2. Radiation Cooling 124
g. Theoretical Study of the Size Distribution of Ice-Fop 129
Crystals
h. Effects of Lower Air Temperature 130
i. Conclusions on the Mechanism of Ice-Fog Formation 132
9. SYNOPTIC METEOROLOGICAL STUDIES OF ICE FOG 135
a. Winter Pressure Systems and Ice Fog in Fairbanks, Alaska 135
b. Analysis of Air Mass Trajectories 136
c. Practical Ice Fog Prediction 137
10. VISUAL RANGE IN ICE FOG 139
11. ELECTRIC PROPERTIES OF j "CE-FOG CRYSTALS 140
a. Experiments with Ice Fog Crystals in Electric Fields 141
i. Non-Uniform Electric Field 141
ii. Uniform Electric Field 144
b. Discussion 147
REFERENCES 154
-------
TABLE OF CONTENTS
(Cont'd)
Page
APPENDIX-THEORETICAL STUDY OF ICE FOG SIZE DISTRIBUTION 159
ABSTRACT 160
1. INTRODUCTION 161
2. THEORETICAL CONSIDERATIONS 163
3. COMPARISON WITH EXPERIMENT 169
4. CONCLUSION 174
ACKNOWLEDGEMENTS 176
REFERENCES 177
vi
-------
PREFACE AND ACKNOWLEDGEMENTS
In response to the rapidly increasing severity of the problem
presented by ice fog in Fairbanks due to the recent increase in city
activities and air traffic, the present research on ice-fog problems
has been supported by the National Center for Air Pollution Control,
Public Health Service, Department of Health, Education, and Welfare under
Grant AP 00449, National Science Foundation, Grant GA-19475 and initially
by State of Alaska funds. The project was also supported in part, by the
Department of Interior, Water Resources Research funds.
This'report combines resea;eh done by the following persona:
Takeshi Ohtake: Principal Investigator
Sue Ann Bowling
Paul J. Huffman
George F. Lindholm
Teizi Henmi
Rudolf Suchannek
Carl S. Benson Consultant
Yosio Suzuki Consultant
Yoshiaki Toda Consultant
Gaishi Onishi Consultant
The researchers are indebted to Dr. Keith B. Mather and many staff
members of the Geophysical Institute at the University of Alaska for their
encouragement, to the above consultants and to Drs. Kenji Isono and
Makoto Komabayasi of Nagoya University, Dr. Thomas E. Osterkamp of the
vii
-------
University of Alaska, Dr. Norihiko Fukuta of the University of Denver and
Dr. Myron L. Corrin of the Colorado State University for their valuable
discussions and comments, and to Messrs. Hiroshi Haramura and
Akio Iwasaki who were working at the University of Alaska for their assist-
ance in field work. The author wishes to thank Miss Sue Ann Bowling of
the Geophysical Institute, University of Alaska for her assistance in
preparing the manuscript. Also the author wishes to thank the management
of the Municipal Utilities System of Fairbanks and the Industrial Air
Products Company, Fairbanks, Alaska for providing locations and electric
power for our instrumentation, and the meteorological staff at Eielson Air
Force Base and the cooperative agencies at the Fairbanks International Air-
port, especially the Fairbanks Weather Bureau and Pan American World Airways,
for providing various meteorological data as well as locations and electric
power for taking micro-photographs. The Naval Arctic Research Laboratory,
Pt. Barrow, Alaska, also provided support for the collection of data on
condensation nuclei at Pt. Barrow. The researchers thank the Institute of
Arctic Biology at the University of Alaska for giving us an opportunity to
use their cold environmental room. Finally, we are grateful to the National
Geographic Society for permission to reproduce Figure 61 from their book
"Alaska".
viii
-------
LIST OF PUBLICATIONS RESULTING FROM PROJECT
ICE FOG RESEARCH SUPPORTED BY GRANT AP00449
phtake T. : Alaskan Ice Fog, Physics of Snow and Ice, Proc. Intern'l
Conf. Low Temp. Sci., 1966, Sapporo, Vol. 1, 105-118. (1967).
Bowling S. : A Study of Synoptic-scale Meteorological Features Associated
with the Occurrence of Ice Fog in Fairbanks, Alaska. University of
Alaska, Master's Thesis, 141pp. (1967).
Huffman P.J. : Size Distribution of Ice Fog Particles, University of
Alaska, Master's Thesis, 93pp. (1968).
Ohtake T. : Freezing of Water Droplets and Ice Fog Phenomena, Proc. Intern'l
Conf. Cloud Physics, Aug., 1968, Toronto, (1968).
Bowling S., T. Ohtake., and C.S. Benson : Winter Pressure Systems and Ice
Fog in Fairbanks, Alaska, Journal of Applied Meteorology, Vol. 7,
961-968. (1968).
Henmi T. : Some Physical Phenomena Associated with Ice Fog, University of
Alaska, Master'?. Thesis, 90pp. (1969).
Ohtake T. and P.J. Huffman : Visual Range in Ice Fog, Journal of Applied
Meteorology, Vol. 8, 499-501. (1969).
Ohtake T. and R. Suchannek : Electric Properties of Ice Fog Crystals,
Journal of Applied Meteorology, Vol. 9, 289-293. (1970).
Ohtake T. : Unusual Crystals in Ice Fog, Journal of Atmospheric Sciences,
Vol. 27, 509-511. (1970).
Ohtake T. : Studies on Ice Fog, University of Alaska, Geophys. Inst. Report
UAG R-211, 179pp. (1970).
Huffman P.J. and T. Ohtake : Formation and Growth of Ice Fog Particles at
Fairbanks, Alaska, submitted to J.G.R.
LIST OF PUBLICATIONS IN PRESS
t
Ohtake T. : Ice Fog and its Nucleation Process, Proceedings of Conf. on
Cloud Physics, Ft. Collins, Colo., Aug., 1970.
Ohtake T. and G. Lindholm : Electron Microscope Study of Ice Fog Crystals,
Journal of Glaciology.
Ohtake T. and T. Henmi : Ice Fog Crystals and Mechanism of their Formation.
Journal of Glaciology or Quarterly Journal of the Royal Meteorological
Society.
ix
-------
LIST OF TABLES
Tables
1. Mean size, concentration and solid water content vs. temperature
at the MUS site.
2. Data of ice-fog crystals in various locations. Data other than the
first two data were taken by the precipitation method (replicas),
The temperature of stream water at Chena Hot Springs was 35C.
3. Concentration of condensation nuclei in and around the Fairbanks
city and far from the city.
4. Average ice nuclei temperature spectrum.
5. Exposure time for vapor method of replication of ice cr^jtals.
6. Percentage of composition of ice fog nuclei determined by electron
microscope and electron microdiffraction. At the MUS site 1C means
ice crystal nuclei and IF means ice fog crystal nuclei.
7. Mean values of precipitation rates and solid water content of ice
fog (by precipitation method).
8. Comparison of the data on humidities between the method used here
and Assman type hygrometer.
9. Observation of evaporation rate of water from pans in low temperature.
10. Radiative temperature of various objects.
-------
LIST OF FIGURES
Figures
1. The solid circles indicate the locations where counts of condensation
nuclei were made.
2. Map showing locations in the Fairbanks area where data on ice-fog
crystal size distributions, nuclei, humidities etc. were obtained
and ice crystal samples were collected.
3. Comparison of size distributions of ice-fog crystals obtained
simultaneously by two methods at the same location and time.
4. Three stage konimeter used for collecting ice-fog crystals.
5. Microphotographs of ice-fog crystals obtained with the ice-fog crystal
sampler represented in Fig. 4.
6. Typical size distribution of ice-fog crystals in downtown Fairbanks,
at -36.5C. Total water content 0.23 mg 1~1, solid water content
0.12 mg 1 , visibility 180 m. particle concentration 153 p cm , at
1240 AST on 8 December 1968. Peaks A, B and C represent crystals from
car exhausts, open water and heating plants (commercial and residential)
respectively. Heights of the peaks are different from place to place
owing to different moisture and temperature conditions,
7. Size distributions of ice-fog crystals at various locations on
2 January 1969, at temperatures between -47 and -50C, excepting that
of Chena H. S. At Chena Hot Springs, collection was made on
1 January 1969 at a temperature of -45C. All distributions on this
figure used the precipitation method.
8a. Concentrations of condensation nuclei at various places in Alaska
using different expansion ratios.
8b. Concentration of condensation nuclei in terms of percentage of total
nuclei to that of nuclei passing through Millipore filters. IF means
the observations were made in ice fog.
9. Smoke from burning dry wood.
10. Smoke from heating plant.
11. Auto exhaust (English car MG).
12. Auto exhaust (Ford Falcon).
xi
-------
LIST OF FIGURES
(Cont'd)
13. Steam fog nucleus in fall season. The nucleus is presumed to be
a combustion by-product.
14. Steam fog nucleus. The nucleus is identified as a sea salt particle.
15. Construction of sheet mesh for electron microscope examinations of
ice-fog crystals.
16. Size distribution of ice-fog crystals examined under the electron
microscope.
17. Size distribution of ice-fog nuclei examined under the electron
microscope.
18. Ice-fog crystal without nucleus, collected at Chena Hot Sp ings
on 31 December 1968. Temperature was -45C.
19. Ice-fog crystal with many dust particles only inside of the crystal.
The crystal was sampled at the MUS site, on 2 January 1969.
Temperature was -45C.
20. Ice-fog crystal with carbon black as a nucleus. The crystal was
collected near the IAP site on 8 December 1968. The white part
at upper right of picture seems to be a kind of spicule formed
from the crystal.
21. Spherical ice-fog crystal with very small nucleus off center.
The crystal was collected near the IAP site on 21 December 1965.
Temperature was -36.4C.
22. Ice-fog crystal and its nucleus sampled at the Fairbanks International
Airport on 1 January 1966. The nucleus is presumed to be a combustion
by-product. The temperature was -43.2C.
23. A spherical ice-fog crystal with off center nucleus. The nucleus
is presumed to be combustion by-product. The sample was taken
near the Noyes slough in winter of January 1965. The temperature
is unknown.
24a. Nuclei of ice-fog crystal. The nuclei were presumed to be combustion
by-products, which may have been in the gas phase in temperate x^eather.
Original outline of ice crystal also illustrated.
24b. Enlargement of the nuclei of Fig. 24a. The crystal was collected
directly onto collodion film at the IflJS site on 21 February 1966.
The temperature was -32.9C.
xii-
-------
LIST OF FIGURES
(Cont'd)
25. Ice-fog crystal with nucleus at center. The nucleus was also
presumed to be soot. Thfi crystal was sampled at the >IUS site
on 23 December 1965.
26. Ice-Cog crystal with nucleus at the center. The nucleus was
presumed to be a combustion by-product. The sample was taken
near the University power plant on 6 January I9ui>. Temperature
was -39,8C and visibility x
-------
LIST OF
(Cont'd)
37. Relationship between mean diameters of plain columns and
temperatures.
'38, Relationship between m'_-an diameters of skeleton columns und
temperatures.
39. Relationship between raean diameters of irregular shaped crystal;;
and temperatures.
40, Mean size distributions of various shaped ic<-fog crystals for
temperature range -31.0 to -32.9C.
41. Mean size distributions of ice-fog crystals for temperature range of
-35.0 to -36.9C.
42. Mean size distributions of ice-fog crystals in various shapes for
temperature range of -39.0 to -41.OC.
43. Percentage distribution of the shapes of ice-fog crystals.
44. Numbers of ice-fog crystals precipitated versus ambient air
temperatures.
45. Solid water contents and precipitation rates of ice fog.
46. Numbers of ice-fog crystals precipitated versus supersaturations
over ice.
47. Ic -fog crystals at Chena Hot Springs, on 31 December 1968.
48. A polyhedral ice crystal found in ice fog at the temperature
of -47C. The crystal was collected at the ?-UJS site on 3 January
1969. Three pictures show the crystal with different focusings.
This crystal may be 14- or 20-faced polyhedral crystal. The
crystal was suspended in silicone oil.
49. A polyhedral crystal looking from the exact top. The crystal was
collected under the same conditon as of Fig, 48.
50, Drawings of the crystals in Figs. 48 and 49-
51. Another polyhedral ice crystal in ice fog under some condition as
of Fig. 48.
Xlv
-------
LIST OF FIGURES
(Cont'd)
52. An ice crystal similar to above polyhedral crystals under sam-
conditions.
53. Ice-fog crystals collect -d under same conditions as Fig. 48.
Mark A shows a thin plate with two half hexagonal thin plater,.
Marks B illustrate "block ice crystal". CL is a plate crysta1
which has another plate angled 60° or 90°.
54. Tee-fog crystals taken under same condition as above. C, shows ,t
half plate and a vertical plate attached with it. The v/eathev
conditions were the same as above.
55. Drawings of th^ crystals A and C, of Ftsjs, 53 and J4.
56. and 57. Block ice crystals. The crystal in Fig. 56 was collect;-.'.;
under the same conditions as Fif*. 43. The. crystals of Fig. 57
were sampled on 23 December 1965 at the MUS site with temperature
of -4.QC ..... The crystals in Fig. 57 were ..suspended .in. fonnvar liouid.
58. Relationship between relative humidities provided by hair
and temperatures during the winter of 1967 to 196 ' at various site-.
59. Arrangement of apparatus for humidity measurement.
60. Water vapor density versus temperatures under ice-fog conditions
at the MUS site. The curves for water- and ice-saturation arc-
according to Smithsonian Meteorological Table.
61. An aerial photograph of Fairbanks taken just before ire fog
forms. (Photo by Mobley, National Geographic Society).
62. Relationship between temperatures and visibilities at the.
Fairbanks International Airport during December 1964 tliroug'i
February 1965.
63. Relationship between temperatures and visibilities at the
Fairbanks International Airport during December 1968 through
February 1969. Compare with the situation as of 4 years ago.
64. Temperature profiles above open water under the conditions of
several ambient air temperatures.
65, Result of calculation of time for water droplet to cool fvon
OC to -30C through conduction cooling or radiation cooling.
-------
LIST OF FIGURES
(Cont'd)
66. A small impactor to check whether the water droplets have been
frozen or not.
67. The parallel plates to produce a uniform electric field.
68. Drawings for the possible explanations of ice-fog crystal
depositions on the parallel wires. The upper drawing (a)
shows ice crystals which have induced dipole moments,
(b) shows crystal which has dipole layer on the surface and
(c) shows force exerted to the ice crystals near positive
and negative electrodes.
APPENDIX
LIST OF FIGURES
Figure
1. Typical examples of temperature of exhaust gases computed from
equation (14). Curve A: automobile exhaust, a = 5 x 10 cm
deg , b = 66.7 cm" , v = 2000.cm sec"-'-- Curve B: exhaust from
heating plant, a = 2 x 10"^ cm deg'1, b = 667 cm"1. VQ =200 cm
sec - Curve C: above open water, a = 5 x 10""^ cm deg ,
b = 6670 cm , v = 20 cm sec""1.
2. Saturation ratio versus'time for cooling rates of Fig. 1.
3. Computed size distributions for ice fog particles produced by cooling
rates shown in Fig. 1.
xvi
-------
!• INTRODUCTION
In arctic and subarctic continental cities, when the temperatures
go down to about -30C, a sort of fog appears. From theoretical consider-
ations, th-'s fog has been believed to be composed of many iiny ice
crystals suspended in the air. This is ice fog, ar-1 it becomes in-
creasingly dense as the air temperature continues to decrease. Since
places away from cities do not normally have ice fog (e-^ept along
heavily travelled roads), the ice fogs have been assumed to be associated
with human activities such as home heating, car exhaust and power plant
exhaust.
Ice fog causes many traffic an-i health problems to inhabitants, as
well as hampering airport activities because of seriously reduced
visibility. The recent discovery of oil in the north slope area of
Alaska has been accompanied by more combustion exhaust from airport
activities, more people working, and more ice fog. The ice fog normally
develops only in a shallow layer (about 50 m thick) the exact thickness
varying with the strength of the ground inversion, which may be extremely
strong.
Investigations of ice fog problems, i.e., the causes of ice fog
formation, various meteorological conditions associated with ice fog, and
the ways to minimize ice fog have been carried out by many investigators.
Most of the studies concerned with ice fog were primarily oriented towards
Alaskan or Fairbanks ice fog as a local problem. However, a careful study
would be a significant contribution to physical meteorology as ice-fog
-------
phenomena are similar to those involved in the formation of ice crystals
in the upper air in most parts of the world. Even though many investi-
gators have studied ice fog, there are many aspects of the problem and
much work to be done. The continuing study of ice fog will therefore
provide an excellent approach to the physics of cloud formation, physical
and synoptic meteorology and micrometeorology, as well as contributing
to the practical problem of this type of air pollution during Alaskan
winter conditions.
Recently Weller (1969) summarized the investigations of the Fairbanks
ice fog which have been done and are being continued by many investigators.
He gives an excellent bibliography on the subject. However, the purpose
of Weller1s paper was to give a better understanding of the Fairbanks ice
fog to people who need information on it related to commerical or in-
dustrial purposes, rather than to provide physical information on ice fog
in general or associated fundamental physical problems. As Weller stated,
Benson's (1965) report can still be considered as the most comprehensive
study of the overall problem of ice fog in Fairbanks and in the surrounding
area. Benson reported the Fairbanks ice fog in detail using a different
approach from that of other scientists.
Detailed introductions and references will be given in each chapter
of this report rather than in this general introduction. The main purpose
of the present research has been to clarify or confirm the mechanism of
ice-fog formation. The conclusions are based upon substantial physical
data observed by ourselves throughout the field and laboratory work, and
synoptic and theoretical consideration.
-------
Since Oliver and Oliver (1949) suggested that ice fog might be a
form of air pollution, the Stanford Research Institute (Robinson et al.,
1955), and Benson (1965) have emphasized that ice fog is a kind of air
pollution. In particular, the report entitled "Ice fog: Low Temperature
Air Pollution" by Benson described the ice fog as air pollution in the
forms of water and pollution other than water at Fairbanks and compared
it with the Los Angeles smog, which is probably the most widely known
form of air pollution. Through the observations of concentrations of
condensation nuclei in and around Fairbanks and the comparison of these
values with the extent of ice fog, as well as through his pioneering
study of ice-fog nuclei with an electron microscope Kumai (1964) also
concluded that the ice fog covers the same area as do high concentrations
of combustion by—products, even though he did not use the words "air
pollution". However, he suggested that air pollution or condensation
nuclei have relatively more important roles than water vapor in forming
ice fog. It is a fact that hazardous gas and dust are considerably
increased at the time that ice fog forms in a city, because at such a
time an intense ground inversion normally develops and under such con-
ditions hazardous gas and dust particles will be trapped near the ground
together with water vapor, droplets and ice crystals, so it is necessary
I
to study each component separately.
From these points of view, the main purposes of the present research
were to clarify the following problems: 1) Which is the more critical
ingredient for the formation of ice fog - dust (nuclei) or water,
-------
2) what is the role of the air pollution other than water in ice-fog
formation, and 3) how does water released from combustion processes,
etc. make ice-fog crystals? The answers to these questions would clarify
the nucleation process of ice-fog crystals. Since the ice fog is
essentially the same phenomenon as the condensation trails (the so-called
contrails) produced by airplanes flying in the higher troposphere, the
research on ice fog also provides information about contrails.
During this research in an extremely cold environment, we have to
expect some instrumental difficulties as well as the observers' physical
problems. In the present project, many kinds of measurements such as
the concentrations of ice-fog crystals, ice nuclei, condensation nuclei
and their sizes and shapes etc.» besides measurements of water vapor
contents were made using several types of equipment and special techniques.
-------
NCHORAGE
KENAI
Fig. 1. The solid circles indicate the locations where the counts of condensation
nuclei were made.
-------
MOVES
SLOUGH
UNIVERSITY
OF ALASKA
UNIV. PLANT
— ESTER DOME
FAIRVIEW
LATHROP
FAIRBANKS
FOODLAND
AIRPORT
EIELSON AFB
(35 KM)
Fig. 2. Map showing locations in the Fairbanks area where the data on ice-
fog^ crystal size distributions, nuclei, humidities, etc. were obtained and ice
crystal samples were collected.
-------
2. MEASUREMENTS OF ICE-FOG CRYSTALS
Thuman and Robinson (1954a) made the first observations of size
distribution of ice-fog crystals at Eielson Air Force Base, 25 miles
southeast of Fairbanks, Alaska. They found that the mean particle
diameter of "droxtals*" (which was the predominant crystal type in ice
fog at -44C) was 13p at a temperature of -40C and the size increased
with increasing temperature to 16y at a temperature of -30C, although
they did not show a complete size distribution in the paper. Later,
Kumai (1964) showed size distributions of ice-fog crystals in terms of
numbers of precipitated crystals including the spherical crystals (which
are the same as "droxtals") with their sizes of 4 to 12y (the peak was
at 7y) diameter for -39 and -41C in downtown Fairbanks and at the Airport,
respectively. From these distributions Kumai figured out that the
-3 -3
concentrations of crystals are 155 crystals cm and 96 crystals cm
_3
and that the solid water contents are 0.07 and 0.02 gm m for the
temperatures of -39 and -41C, respectively.
These size distribution studies carried out by Thuman and Robinson
and by Kumai used a precipitation collection method in which particles
are collected on a slide glass by gravitational precipitation or
sedimentation at terminal velocity. Even though the precipitation method
(
gives very consistent results, it normally produces an uncertainty in
A droxtal, which is the term used by Thuman and Robinson, is a small-
sized poorly-formed ice crystal. It is an equant solid particle with
rudimentary crystal faces, and thus seems to have characteristics of
both droplets and crystals, according to them.
-------
concentration of such particles from the precipitation method, because
we have to use the Stokes1 fall speed for each size particle. However
according to Ohtake (1964), small air turbulence disturbs such fall speeds
and may give a lack or excess of smaller particles. The concentration of
crystals is derived by dividing the numbers of crystals precipitated on
a glass by Stokes1 fall speed for corresponding sizes. The fall speeds
for the smallest sized crystals are much smaller, for instance 0.1 mm
sec for 2y diameter ice crystal, than the actual possible turbulent
speeds. So the results from the precipitation method give variable
concentrations, especially for less than lOy diameters. Also Benson
(1965) stated that updrafts in ice fog in the downtown Fairbanks area
were likely to be greater than the falling velocit> of the small particles.
The error in size distributions introduced by turbulence or updrafts is
demonstrated by Fig. 3, which shows the results obtained at the same time
and location by a) the precipitation method, and b) the impaction method
to be discussed below.
Considering these disadvantages of the precipitation method, we used
the impaction method to obtain size distributions of ice-fog crystals.
The method is similar to that used by Kozima et al. (1953) for collecting
sea-fog drops. Figure 4 illustrates the three stage impactor, improved
by K. Kikuchi (unpublished) of Hokkaido University, which we used. For
the collection of ice-fog crystals we used only one of the three stages.
As a measured volume of air (V) is drawn into the nozzle (Q) by rapidly
pulling out the plunger (P), ice-fog crystals contained in the air sanple
are deposited in a thin film of silicone oil on the small microscope slide
-------
30
LU
CM
_ .
o —
o: >-
I- cr
2 o
o
o
10
0
IMPACT ION
- PRECIPITATION
10 20
DIAMETERS
30
Fig. 3.
Comparison of size distributions of ice-fog crystals obtained simul-
taneously by two methods at the same location and time.
-------
Q
Fig. 4. Three stage konitneter used for collecting ice-fog crystals.
-------
(S) (1.8 cm square). The area onto which the crystals are deposited
depends on the nozzle diameter (Q), separation (y) between nozzle and
slide, and the rate at which the plunger is pulled out. These parameters
must be adjusted so that all particles are collected and are in the field
of view of the microscope, or are concentrated at the center of the slide
glass. This means that collection efficiency is 1.00 if the smallest
crystals were circularly scattered around most of the other crystals
deposited at the center. Two pictures are shown in Fig. 5 as examples.
Some ice-fog crystals were falling onto the slide during the time the
picture of the slide was being taken, but this introduces a negligible
error. The values we used were Q = 1.0 mm and y = 2.5 mm. The volume
of air sampled V was chosen according to ice fog density to give the
largest possible numbers of crystals without overlapping, normally 8 c.c.
The only disadvantage of this method is that any observer must practice
to obtain good results in preventing overlapping without letting the
smallest sized crystals escape from the slide. The difficulties are to
choose an appropriate suction speed and volume as well as to select the
correct thickness and kind of oil film on the slide.
An automatic sampler which was made and operated by Huffman (1968),
who was a member of this project, was working well as long as he operated
it. It needs some improvement. For further information about this refer
to Huffman's thesis (1968).
Nevertheless we sometimes used the precipitation method to determine
size distributions of ice-fog crystals, because this method gives con-
sistent data by a very simple technique even though some errors in the
11
-------
IOO/C6
Fig. 5. Micro photographs of ice-fog crystals obtained with the
ice-fog crystal sampler represented in Fig. 4.
-------
smallest crystals are expected. Another advantage of this method is that
it is possible to obtain formvar replicas of ice crystals or water drop-
lets on a glass slide very easily, while the impaction of crystals re-
sults in bouncing and the escape of crystals from a dry slide glass coated
by formvar, even though it does not occur on a slide coated by oil.
Procedures for making replicas of ice-fog crystals will be mentioned in
a later chapter.
On the basis of the experimental data obtained, the following con-
clusions can be reached regarding the concentrations, size distribution,
and solid water content of ice-fog crystals:
1. Mean diameters change from place to place, which implies that they
depend upon temperature, humidity and the local supply rate of ice-fog
crystals. Even at the same place, an increase of population, traffic,
etc. produces thicker ice fog (more crystals per unit volume of air) and
larger solid water contents, but normally the size distribution stays
constant if the source(s) of the ice fog does not change.
2. During the winters of 1965 through 1969 the mean diameter of crystals
in air traffic areas away from downtown Fairbanks (i.e. Fairbanks Inter-
national Airport and Eielson Air Force Base) was about 5y, in the downtown
area (MUS) it was 3 to 4y, and at the places adjacent to open water such
as the Industrial Air Products parting lot (along the Chena River down-
stream from the MUS power plant), Chena Hot Springs, and Manley Hot
Springs, the diameters of crystals were about lOy. All values are for a
temperature around -40C or a little colder.
13
-------
3. The distribution is usually bimodal or trimodal in downtown Fairbanks
and unimodal at the other locations. The typical distribution in the
downtown Fairbanks area is shrwn in Fig. 6.*
4. The concentration of ir.e-fog crystals also changes from place to place,
and from time to time. For instance, the concentration is very high in
the middle of downtown Fairbanks and almost zero outside of the city.
The mean value of concentration of ice-fog crystals was about 150 part-
-3
icles cm which roughly agrees with Kumai's (1964) data although he
observed the concentration only twice. The maximum value we got using
_0
the impaction method through 1968 was 227 p cm and it is predicted tnat
the maximum value would increase with lov/ering of temperature and with more
increase in human activities in the town.
5. Concentration of ice-fog crystals should depend not only on temper-
atures but also on the supply rate of moisture to the air (or incoming
and out-going water droplets or ice-fog crystals). The vapor supply is
more important than the temperature at temperatures lower than -35C. So
observations of the temperature dependence of the concentration were made
at definite locations (with nearly fixed conditions of moisture supply
rate), but at different temperatures. The variation of concentration with
temperature is based on observations using the precipitation method at the
MUS site during the winters of 1967 through 1969.
The datum of solid water content shown on page 26 of Weller's report
(1969) (supplied from the author's result) was a misprint. It should
be changed to "Total Water Content" and solid water content should have
been 0.12 gm m ; crystal concentration was 153 p cm .
14
-------
40
C\J
O
x.
CO
30
o
I-
cr
I 20
<
LU
o
o
o
10
0
10 20
DIAMETER (p.)
30
Fig. 6. Typical size distribution of ice-fog crystals in downtown
Fairbanks, at -36. 5C. Total water content 0. 23 mg I"1, solid
water content 0.12 mg 1~^, visibility 180 m, particle concentration
153 p cm ~3, at 1240 AST on 8 December 1968. Peaks A, B and
C represent crystals from car exhausts, open water and heating
plants (commercial and residential) respectively. Heights of the
peaks are different from place to place owing to different mois-
ture and temperature conditions.
-------
TA3SLE 1
Mean Sine, Concentration and Solid Water Content
vs. Temperature at the MUS Site
Temperature Mean Diameter Concentration Solid water „
^C M crystals cm" content ntm m
-47 3 715 0.05
-40 8 250 0.09
-35 22 9 0.05
-30 33 1 0.02
Despite the fact that the absolute numbers of concentration of ice-fog
crystals may not be very accurate, because the data were derived from the
precipitation method, the temperature dependence of the concentration is
very clear in the table. The dependence of the solid water content on
temperature was not clear.
From December 30, 1968, through January 6, 1969, the Alaskan interior
experienced extremely severe cold weather with ice fog. During this time
we obtained data on ice-fog crystals at various locations. Most of the
data were obtained through the use of replicas of crystals collected by
precipitation onto glass microscope slides. Table 2 shows the concent-
rations and solid water contents of ice-fog crystals and Fig. 7 shows
the size distributions at all sites. In both the table and figure all
samples except those of December 30, 1968, were collected by use of the
precipitation method on slides as replicas to minimize the time spent.
Even though the values of concentration and size distribution are not
entirely accurate because the precipitation method was used, comparison
of the data at several sites may be valuable. In the concentration data
16
-------
CO
500 r
CXI
ro'
5
O
\
a: 400
LU
CD
oc
o
O
o
L_
I
LJ
O
O
H-
<
a:
UJ
o
z
o
o
300
20°
100
0
—°— CHENA H. S.
LATHROP HIGH SCHOOL
FAIRVIEW MANOR
MUS
IAP
AIRPORT
5 10
CRYSTAL SIZE (p.)
20
Fig. 7. Size distributions of ice-fog crystals at various locations
on 2 January 1969, at temperatures between -47 and -50C, excep-
ting that of Chena H. S. At Chena H. S, collection was made on 1
January 1969 at a temperature of -45C. All distributions on this
figure used the precipitation method.
-------
TABLE 2
Data of ice-fog crystals in various locations. Data other than the first two data were
taken by the precipitation method (replicas). The temperature of stream water at Chena
Hot Springs was 35C.
Date
Dec. 30, 1968
Dec. 30, 1968
Dec. 31, 1968
Dec. 31, 1968
Jan. 1, 1969
Jan. 1, 1969
Jan. 1, 1969
Jan. 1, 1969
Jan. 2, 1969
Jan. 2, 1969
Jan. 2, 1969
Jan. 2, 1969
Jan. 2, 1969
Jan. 2, 1969
Jan. 2, 1969
Time exposed
(AST)
12:15(suction).
12: 47 (suction)
20:38-20:44
21:12-21:19
10:44-10:54
11:00-11:05
11:12-11:17
11:22-11:32
11:14-11:25
11:38-11:43
11:54-12:04
13:22-13:30
13:51-14:17
14:37-14:46
14:58-15:08
Location
MUS
MUS
Chena Hot Springs
Chena Hot Springs
Chena HS (near stream)
Chena Hot Springs
Chena HS (near stream)
Chena HS (50m from ")
MUS
IAP
Fairview Manor Apt.
Lathrop High School
Int'l Airport
College residential area
University campus
remperature
°C
-38.0
-39.0
-43.9
-42.2
-45.2
-44.8
-44.0
-44.5
-47.0
-48.0
-47.2
-47.0
-50.0
-48.5
-48.5
Visibility
m
60
90
?
?
-
-
-
—
60
70
60
60
120
100-300
80
Concentration
>2.5y <235y
p. cm"
Solid Water
Content
gm m~
210 (all together)! 0.064
227 ( " )
266 39
724 32
1780 465
373 506
1231 691
77 68
715 889
549 136
163 660
260 1479
378 403
185 25
350 546
0.131
0.073
0.179
0.127
0.042
0.074
0.022
0.055
0.036
0.012
0.020
0.021
0.009
0.019
00
-------
in Table 2, crystals smaller than 2.5y were too small to be counted
accurately in the optical micro-photographs, but are listed separately
to provide some possible findings.
Under similar conditions, both concentration and solid x
-------
3. MEASUREMENTS OF CONDENSATION NUCLEI
Many reports have been published of condensation nuclei counters
using various degrees of supersaturation. Especially, Kocmond (1965) and
Radke and Hobbs (1969) tried to observe condensation nuclei active at the
low supersaturations which occur in cloud formation associated with updrafts
in the free atmosphere. This is the most desirable way to observe the real
concentration of condensation nuclei for the natural clouds. Normally less
than 1 percent of supersaturation is expected in nature (Houghton, 1951)
and sometimes even when the humidity is only 75 to 80 percent, in the cities
along the ocean coast, we can easily find droplets as haze or smog
(Yamamoto and Ohtake, 1955). However, as will be described later, in the
case of ice fog we have much higher supersaturations than the above values
at the source of the fog. The supersaturation is approximately 300 to 400
percent or more in the vicinity of ice-fog sources.
Because of this, we used a traditional condensation nuclei counter,
the Gardner Small Particle Detector, which counts Aitken nuclei (measur-
—7 —5
able size ranges are larger than 1 x 10 or larger than 1.3 x 10 cm)
and which has supersaturations of 300% to 400%. The values provided by
the counter are not useful for natural clouds and normal fogs but valid
for condensation nuclei for the fog near open water and exhausts because
of similar degrees of supersaturation.
Table 3 shows the data we obtained in and around the city of Fairbanks
and other locations including Manley Hot Springs, Chena Hot Springs,
20
-------
Pt. Barrow, Anchorage, Juneau, etc. Most of these observed values are
similar to Landsberg's results (1938), but the data in Fairbanks
(essentially same as "town" in Landsberg's data) are higher, especially
in the ice-fog season. The observed maximum concentration of condensation
f\ —^
nuclei was 6 x 10 p cm in the middle of the downtown area. This value
is much higher than the absolute maximum value for towns in Landsberg's
data. Also the values at Manley, Chena and Big Delta were nearly the
same as the absolute minimum values for country inland data.
Sometimes, to get a size spectrum of the nuclei, several kinds of
Millipore filters such as 14, 8, 5, and 0.45y pore sized filters were
used. However, more than 98 percent of the total nuclei (larger than
1.3 x 10 cm in radius) were captured by the 8y filter. Figure 8b shows
the ratios of the concentration without filters to that with filters.
From the figure it can be said that the nuclei at the MUS site consisted
of smaller particles than at the other sites. These nuclei must come from
automobile exhausts. Using smaller expansion ratios of 15 inch Hg and 5
inch Hg of the instruments pressure gauge reading, instead of 26 inch Hg
(the value of pressure suggested by the maker to measure nuclei larger than
1.3 x 10~ cm radius), gave ratios similar to those of Fig. 8b. Figure
8a shows that the nuclei in areas of heavy traffic decrease in number
I
with decreasing expansion ratios, while the nuclei in clean air areas
such as Circle, Manley and at sea shore sites such as Kenai and Anchorage
do not change much with changing pressure gauge readings. Unfortunately,
the absolute values of the expansion ratio is not known for these data.
However, these data show the same tendency as the Millipore filter method.
21
-------
UJ
_i
o
CO
z
UJ
Q
z
o
o
CO
-------
100
90
80
70
UJ
o
or
UJ
a.
60
UJ
50
40
30
20
10
WITHOUT
FILTER
MUS
(IF)
FOODLAND
(IF)
WITHOUT IF
ON JAN. 22, 1966
8
14
PORE SIZE (p.)
Fig. 8b. Concentration of condensation nuclei in terms of percentage
of total nuclei to that of nuclei passing through Millipore filters. IF
means the observations were made in ice fog.
-------
Obviously, the observed concentrations of condensation nuclei shown
in Table 3 are much greater than those of ice-fog crystals even under
-50C conditions. These observations imply that only a very small fraction
of the condensation nuclei act as real ice-fog nuclei, and the remaining
large nun.jers of condensation nuclei are floating in the air co-existing
with the ice-fog crystals. Huge numbers of dry, unactivated nuclei or
dust particles including lead compounds resulting from car exhaust as
well as carbon monoxide, carbon dioxide, gaseous sulfur compounds and
other gases hazardous for our health are also associated with ice-fog
crystals.
Another interesting point is that even though the concentration of
_3
condensation nuclei is only about 300 p. cm at Chena Hot Springs under
ice-fog conditions, we found about twice that concentration of ice-fog
crystals. From the electron microscope study, which will be mentioned
in more detail in a later chapter, we found many ice-fog crystals without
any nuclei. This means that a high concentration of condensation nuclei
is not a very important factor for the formation of steam fog or the
final ice fog when temperatures are lower than about -40C, although it
is an indication of the action of combustion which produces water vapor
and condensation nuclei or dust and gases.
24
-------
4. MEASUREMENT OF ICE NUCLEI
The definition of an ice nucleus is, according to Wegener (1911),
a nucleus necessary to form an ice crystal in the atmosphere; Wegener
assumed t^at a sublimation nucleus was involved. However, Fournier
d'Albe (1949) and Weickmann (1949) made some experiments and both of
them concluded that ice crystals can be formed only under conditions of
water saturation or supersaturation over water, and they suggested that
a thin water film might first condense on the surface of a solid particle,
and the film would then freeze. From this point of view they called the
ice nucleus a freezing nucleus. Houghton (1951) criticized their term
of freezing nucleus and stated that there is certainly a clear physical
difference between ice crystals formed in such a way and those which are
formed by the freezing of a liquid drop which has already attained cloud
drop size. In the latter case it is evident that a freezing nucleus is
involved. Until more information is available it would seem preferable
to call all nuclei on which ice forms below water-saturation sublimation
nuclei. Most recently, Isono (1969) suggested that an ice nucleus is a
particle which accelerates the formation of an ice crystal from water
vapors regardless of the exact process of ice crystal formation. This
definition includes both sublimation nuclei and freezing nuclei as de-
t
fined by Fournier d'Albe and Weickmann. A freezing nucleus is then a
particle which acclerates the formation of an ice crystal in a water
droplet. These definitions seem to be reasonable. Analogously, a con-
densation nucleus is a particle which accelerates the formation of a
water droplet in air.
25
-------
Sometimes an ice nucleus or freezing nucleus is confused with a
condensation nucleus especially in the study of the ice fog. However,
an active condensation nucleus such as NaCl is not effective as an
active freezing nucleus, at least at temperatures higher than -15C.
And active ice nuclei such as Agl are not active condensation nuclei.
Under ice-fog conditions, since the temperature is very low, any kind
of particles contained in water drops may accelerate freezing of the
drops at a higher temperature than would be observed for pure water drops.
To form a water fog some condensation nuclei are needed under normal
atmospheric conditions. These nuclei, after acting as condensation
nuclei, may act as freezing nuclei at lower temperatures. On the other
hand, ice nuclei may form ice-fog crystals even under sub-water saturation
conditions. Up to the date, no measurement of ice nuclei at effective
temperatures lower than -30C has been published.
A temperature spectrum of ice nuclei concentration was obtained at
Ester Dome using the same diffusion-type ice nuclei counter with a sugar
solution detector which had been used by the author at Tohoku University,
Japan (Ohtake and Isaka, 1964), as can be seen in Table 4. In the author's
experience (unpublished) and also that of others (e.g. Fletcher, 1966),
the concentration of ice nuclei in the atmosphere rises an order of mag-
nitude with every 4C decrease in temperature. Applying this temperature
dependence to the concentrations at Ester Dome for a temperature of -21C,
—1 —3
we derive 25,000 nuclei H (or 25 cm ) of ice nuclei concentration at
effective temperature of -35C.
26
-------
TABLE 3
Concentration of Condensation Nuclei in and around
the Fairbanks city and far from the. city
Juneau Airport
Mendenhall Glacier
Juneau City
Douglas Marine Station
Clear Air Force Base
Ester Dome
University of Alaska
Airport - Fairbanks
Eielson Air Force Base
Base Operation
Cooling Pond
Power Plant
Airport Road w/water fog
Richardson Highway
Black Rapid Glacier
Anchorage (sea shore)
Anchorage (city)
Circle Hot Springs
Manley Hot Springs
Chena Hot Springs
Chena Hot Springs Road
Nuclei cm
26 inch Hg
without filter
750
1,500
160,000
14,000
2,500
500^5,000
20,000^60,000
5,000
12,000
30,000
70,000
2,000^-100,000
500
18,000
20,000
1,300
200-vl,000
300^700
200
Date
December 2, 1965
December 2, 1965
December 2, 1965
December 2, 1965
June 18, 1965
June 21, 1965 (and
many times)
June 21, 1965 (and
many times)
June 21, 1965
July 30, 1965
July 30, 1965
July 30, 1965
June 21, 1965
August 11, 1965
August 11, 1965
August 23, 1965
August 14, 1965
August 17, 1965
December 31, 1968
January 1, 1969
Size Range
r>1.3 x 10~5cm
r>1.3 x 10~5cm
r-1.3 x 10"5cm
r>1.3 x 10~5crr.
r>l x 10 cm
j-.. 1 .. 1° C~.
r>l x 10~ cm
r>l x 10 cm
r>1.3 x 10 cm
r>1.3 x 10~5cm
r^l. 3 x 10 'cm
r>1.3 x 10 cm
r>1.3 x 10~5cm
r--1.3 x 10~ cm
r>1.3 x 10 cm
r>1.3 x 10~ cm
r>1.3 x 10 cm
r>1.3 x 10 cm
r>1.3 x 10 cm
27
-------
TABLE 3
(Cont'd)
'
Valdez
Kenai
1 IloiTier
; Uirch Hill
| Hani 1 ton Acres, Fairbanks
ML'S
TAP
2nd Street
Foodland
Big Delta(near river)
Nuclei cm
26 inch Hg
without filter
40,000
700
5,000
700
1,500
120,000
50,OOCM80,000
6,000,000 (may")
26,000-^100,000
5,000,000
500,000'
200^500
Date
August 30, 1965
August 29, 1965
August 26, 1965
November 8, 1965
October 29, 1967
October 29, 1967
(and many times)
December 4, 1967
(and manv times)
(many times)
December 4, 1967
January 2, 1966
January 2, 1966
Size
r>1.3
r>1.3
r>1.3
r>1.3
r>1.3
w/l.F.
r>1.3
r>1.3
w/T.F.
w/l.F.
- no I
but
r>1.3
Ran Re
:; 10~5cm
x 10 cm
x 10~ err.
x 10~5cm
x 10"5cn,
x 10 cm
r 1.3
r 1.3 x 10~ i-.n
. F. along 11 ifh
I.F. above rl\v
x 10 cm
28
-------
ice
•.•(-T>:-.ra!'.»irc ~~--T U
T-S pe 01 Counters
!
Diff union
Acouiticfl
Expansion
— i ! !
0.6
—
1
(
- - ! « i
5
—
*•
"*. '
7
0.3
0.3
•!i->n«,r
-'' '
0^
__
••
•it'.irt-
-.: -.
O
2
6
25
- '.•' ;
ion
3^0
\
.,,,-Vi il ~1
-1
nurl f- 1 2,
—i
However this value v;as nol. rV.i/uriimir.1 by direct mcr.rji.irc'Moiit-
Aftcr obtaining an 1JGAR Eiooiip'-ic.-i? Ice riiiclcl counter (Longer» o.t -il . ,
1967), we also obtained the te;:[.-:r-Uurc Sj.itiCi:L'uin °f "-'-CG nuclei con-
centration with it, as shown in Tublc 4. From the Table we see. 100
nuclei Si for -35C, which is a much smaller concentration than thcit
of actual ice-fog crystals. Also x^e recognized miss-counting of ice
crystals formed in the chamber of the counter due to smaller size of ice
crystals than could be detected v?ith tlic acoustical sensor of the counter.
This appears to be a problem with this machine at temperatures lower
than -25C.
An expansion type counter which is called the Bigg-Warner counter
(Warner, 1957) and was developed by ESSA (Kline and Brier, 1961) was
available at Colorado State University. Using this, another temperature
spectrum of ice nuclei was measured at the HAD station at Climax, Colorado
As can be seen in Table 4, 300 n. 9, for -35C was obtained. Even though
the detection of ice crystals by this counter seems to be much better thar
29
-------
by the NCAR acoustical counter, this number is still a very UKusll ruunber
compared to the concentration of the ice-fog crystals. Also 300 n.Ł.
is the maximum detectable number by the expansion type counter. The
concentrations of ice nuclei for various effective temperatures varied
by a factor of 10 per 4C in the temperature range between -20C and -35C
in the case of the expansion chamber. Assuming that this relation is
valid to -39C, we may expect 3,000 n. I for a temperature of -39C,
which is still almost negligible in comparison to observed ice crystal
concentrations in ice fog at a temperature of -39C.
..owever, these observations were made at places where the air is
relatively clean (where condensation nuclei numbers are 200 to 2,000
n.cm ) and it is expected that the air which contains a lot of conden-
sation nuclei might produce many more ice crystals in the expansion
chamber, because the condensation nuclei which made water droplets in the
chamber may act as freezing nuclei in the water droplets under the lower
temperature conditions. In fact, adding some combiistion by—products to
the air to be tested before the expansion obviously resulted in the form-
ation of many more ice crystals in the chamber under the same temperature
conditions. In the chamber of the expansion counter, the degree of super-
saturation appears to be very high, even though only momentarily. Thus
it is suggested such condensation nuclei are very important for the
formation of many ice crystals in high supersaturation environment in the
temperature range between Ł -20C and -v -37C. In other words, many con-
densation nuclei in the air must act as freezing nuclei under ice-fog,
conditions.
30
-------
In the air at temperatures lower than -37C (diameter of water drop-
lets is assumed about lOy) some droplets would freeze spontaneously.
In clean air with a very high moisture content, combination of homo-
geneous condensation and spontaneous freezing must result in a threshold
temperature between -37 and -40C. In polluted air in the same temperature
range, such a threshold would not be recognized because the crystals can
be formed by both homogeneous and heterogeneous nucleation on pure and
polluted water droplets.
31
-------
5. RELATIONSHIP BETWEEN CONDENSATION AND ICE NUCLEI
AND ICE-FOG CRYSTALS IN THEIR CONCENTRATIONS
Although we obtained some data on the concentrations of ice nuclei
in the air at a temperature of -35C, we had better compare the concent-
ration of ice-fog crystals to that of condensation nuclei (including
nuclei acting as freezing nuclei), as discussed above. We can assume
maximum value for the concentration of the condensation nuclei of
5 x 10 p cm in the downtown area* and 10 p cm in the outlying areas
during ice-fog conditions, while the concentration of ice-fog Crystals
_3
at a temperature of about -35C in downtown Fairbanks is arouao 200 p Cm ,
-3
at most 1000 p cm .As was mentioned before, the temperature inversion
during ice fog keeps all water vapor, solid dust, gases, and possibly
liquid aerosols as well as cold air near the ground.
There is then no problem in finding sufficient nuclei for the
observed ice-crystal concentration in the air of Fairbanks at low
temperatures, especially temperatures below -37C. In fact it seems that
only 1/1000 of the dust particles in the city act as nuclei regardless
of whether the nuclei are condensation or freezing-nuclei. We may ask
what happens to the remaining 999/1000 of the dust particles which may
be considered air pollution in the normal meaning. Apparently these dust
particles are co-existing with ice-fog crystals. Benson (1965) stated
This number will increase with the further development of human activity,
and may become a serious problem for human health. The city council or
members of city development planning group are urged to consider seriously
such high contents of dust as well as hazardous gases, not only for the
effect on ice fog directly, but also for the general health of the
residents.
32
-------
"The small Ice crystals which make up Ice fog are often associated with
impurities. This is especially true of those formed from combustion
products." He collected ice fog and snow crystals precipitated on poly-
ethylene sheets at various places in and around Fairbanks. After melting
those into water, he measured electrical conductance and also filtered
particles suspended in the melted water precipitated. In these obser-
vations he found much heavily polluted water at the center of the city,
while the outlying area was clear of pollution. He pointed out that
"the electric conductance is, of course, due to free ions in tne melt
water." Furthermore he stated "They could be associated with the crystal
nuclei which are found in each crystal, or they could have been con-
centrated on the crystal surfaces." However, according to our observations
of concentration of dust particles (or condensation nuclei) using 14y
pore sized Millipore filters to remove all ice-fog crystals and larger
5 3
dust particles, we still had as many as about 10 p dust particles per cm .
This shows that huge numbers of particles are still co-existing with the
ice-fog crystals. Also from the electron microscope study of ice-fog
crystals we could not find many crystals which had much dust either on the
inside or their surfaces. In other words, such dust or pollution must
have been precipitated on the polyethylene sheets independently of ice-
fog crystals.
33
-------
6. ELECTRON MICROSCOPE STUDIES OF STEAM-FOG
AND ICE-FOG CRYSTALS AND THEIR NUCLEI
Kumai (1964) was the first to try an electron microscope study of
ice-fog nuclei in the Fairbanks area. The main purpose of an electron
microscope study of such fog or cloud droplets, ice crystals, and snow
crystals is to separate the nuclei of hydrometeor particles from unacti-
vated dust particles which co-exist with the hydrometeors. Historically,
K'dhler (1925) concluded that most cloud nuclei must be sea salt particles.
He collected bulk samples of cloud droplets or rime on a high mountain,
chemically analyzed the collected water and found high concentrations of
NaCl. However, in the electron microscope studies of cloud and fog nuclei
by Kuroiwa (1951), Ogiwara and Okita (1952), and Yamamoto and Ohtake
(1953 and 1955) it has been possible to inspect the nuclei of individual
droplets separately from airborne particles other than water droplets.
This work has led to the conclusion that most cloud or fog nuclei are
very small combustion by-products, although a few large sea salt particles
also have been found. Salinity of bulk water is controlled by the small
numbers of sea salt particles. Measurements of bulk water could not have
shown this. It is tedious work, but electron microscope studies are
essential in the field of nucleation processes.
Kumai (1964 and 1965*) tried this for ice-fog nuclei after his
successful work (1951 and 1963) on snow-crystal nuclei. He reported that
most of the ice-fog nuclei were combustion by-products and 10 to 21%
were clay particles. He also reported that all the ice-fog crystals he
*
Co-authored by O'Brien.
34
-------
collected had nuclei. However, in order to inspect such very small nuclei
as residues of micron size ice crystals which were evaporated without
melting, we must be certain we have a very clean film as the ice crystal
support. If not, we can not distinguish between contamination due to
poor preparation and true nuclei. Also the technique of identifying
nuclei on the electron microscope screen with crystals on optical photo-
micrographs taken earlier of the same sample grids is difficult and time
consuming even wi,th a special grid having a finder pattern. In the
present project, we tried to use the replication method for ice-fog
crystals in the electron microscope, so that crystals and nuclei were
both visible on the electron micrographs.
a) Smoke Samples:
Some samples of smokes which were considered to be possible sources
of fog nuclei were taken and examined in the electron microscope. The
specimens included car exhausts, smoke from oil - and coal-burning
furnaces, exhaust smoke from a power plant, and smoke of burning dry
grass or dead trees. The electron micrographs for them are shown in
Figs. 9 to 12. Generally speaking, combustion for heating and power
plants produced soot or carbon black only as solid particles, while
car exhausts showed various shapes. Gasoline for cars includes lead
i
compounds and other additives to prevent knocking. Also the characteristics
of the exhaust vary from car to car, and as a function of motor adjustment,
age of the car, etc. Sometimes car exhausts showed a grandular structure as
shown in Fig. 12. Smoke of dead plants showed very thin and semi-transparent
structures on fluorescent screen in electron microscope.
35
-------
Fig. 9. Smoke from burning dry wood.
Fig. 10. Smoke from heating plant.
Fig. 11. Auto exhaust (English car MG). Fig. 12. Auto exhaust (Ford Falcon).
Fig. 13. Steam fog nucleus in fall season. pig. 14. Steam fog nucleus. The
The nucleus is presumed to be a combus- nucleus is identified as a sea salt
tion by-product. particle.
-------
b) Steam-Fog Nuclei:
On clear nights during the summer and fall seasons, steam fog
usually develops due to radiative cooling of the air near the ground
and the constant warm temperature of water surfaces especially near
lakes. The mechanism of formation of steam fogs seems to be similar to
that of ice-fog formation above open water except for absolute temper-
ature. Under such conditions, fog can be seen above and around lakes
or damp ground even in unpopulated areas or areas far from traffic.
Steam-fog droplets were collected on electron-microscope grids and their
nuclei were examined with the electron microscope. Most of the nuclei
were identified as combustion by-products (see Fig. 13 as an example).
In addition to the combustion nuclei, a few sea-salt particles have been
detected as shown in Fig. 14. These nuclei were identified as in the
report by Yamamoto and Ohtake (1953).
c) Ice-Fog Crystals and Their Nuclei:
1. Method of Making Specimens:
To prepare the ice-fog samples for electron-microscope examination,
the vapor replication method was used instead of the liquid method. The
liquid method has two main disadvantages: (i) the ice-fog crystals will
be displaced and concentrated into a small portion of the slide glass
!
because of the surface tension of the liquid, and (ii) the film is
thicker than desired for use with the electron microscope. The vapor
method, which was developed by Schaefer (1962), prevents any displacement
of ice particles and produces good replicas for the electron beam. The
exposure times for a slide glass coated with 0.5% formvar solution to be
37
-------
in contact with chloroform vapor were determined by means of saturation
vapor pressure over chloroform for various temperatures. On this matter,
the only exposure time Schaefer (1962) gives is 20 seconds for -20C. The
exposure times for chloroform vapor at various temperatures are shown in
Table 5.
TABLE 5
Exposure Time for Vapor Method of Replication of Ice Crystals
Ambient Temperature
Exposure Time
(°C)
(sec)
0
8
-10
13
-20
20
-30
40
-40
80
-45
120
-50
160
To prepare a specimen sheet mesh (grid) for the electron microscope,
according to Tanaka and Isono's (1966) method, a collodion film and a
layer of carbon were deposited between a formvar film and metal sheet
mesh (see Fig. 15). The films can be made at room temperature and the
carbon layer is deposited in a vacuum evaporator. These meshes with an
outer layer of formvar were then cooled to ambient temperature and exposed
to the ice fog. After the ice-fog crystals were allowed to deposit on
the grids (which were normally on slides) for several minutes, the mesh
was exposed to chloroform vapor in a closed petri dish for the time shown
in Table 5. The exposure of the formvar film with ice crystals on it to
chloroform vapor allows the film to soften, and because of surface tension
the softened formvar covers the ice crystals. After exposing the film
to the vapor, the slide glass is waved in the air to remove excess chloro-
form vapor from the slide. The slide glasses (or grids) should be kept
cold and dry during the sublimation of the ice crystals after the replica
38
-------
ICE CRYSTALS
FORMVAR FILM - - -
COLLODION FILM - x -CARBON SPATTERED
-SHEET MESH GRID
SLIDE GLASS ]////
Fig. 15. Construction of sheet mesh for electron micro-
scope examinations of ice-fog crystals.
-------
is formed. The specimens made in this way are ready to be examined by
regular electron microscope techniques including chromium shadow casting.
Laboratory tests for permanency of replicas and for migration of nuclei
were made prior to the ice-fog season. Under an optical microscope the
real ice-fog crystals and their replicas deposited on a slide glass in a
specially made glass container were observed during evacuation of the
container. The results of the test were quite satisfactory.
To make a plain slide glass with replica, a simple formvar film can
be coated on a slide glass rather than the three layers of film, and
further procedures will be similar to those for grids. For thicker or
larger crystals an increased thickness of formvar will be needed.
Direct collections of ice-fog crystals on specimen grids coated with
formvar film without replication were also made during the 1965-1966 winter.
An ice-fog crystal or any kind of hydro-particle, such as a steam fog
droplet, will evaporate in an electron microscope because of the high
vacuum inside. So we cannot see the outlines of the original ice-fog
crystals and steam-fog droplets on a specimen grid. Up to date, the
original positions of particles on the grids had been deduced from
pictures taken with an optical microscope. Using these pictures, the
exact positions of the ice-fog crystals or steam-fog droplets, even to
the exact outline of ice particles, could be determined with accuracy
as great as 1 to 2y by means of the "Specimen Position Indicator" built
in the JEM-electron microscope. This method has also been used for
checking the displacement of nuclei or other effects during the re-
plication of ice-fog crystals.
40
-------
Many specimens of ice-fog crystals and ice crystals for electron
microscopic examinations were collected at different places and times.
It was expected that the ice-fog crystals would have different kinds of
nuclei according to their conditions of formation.
2. Size and Shape of Ice-Fog Crystals:
As was mentioned previously, the size distribution and shape of
ice-fog crystals change with variations of temperature. Since ice-fog
crystals have been collected only by the precipitation method for electron
microscopy, the deduced size distributions and average shapes of ice-fog
crystals may be different from the real ones. However, we tried to take
electron micrographs of randomly sampled ice-fog crystals from sheet mesh
grids. Even though the total number of pictures of ice-fog crystals at
each station or total stations is not enough to allow construction of an
accurate distribution, we can find average sizes of ice-fog crystals and
compare them with the average sizes of ice-fog nuclei. Figures 16 and 17
show size distributions of both ice-fog crystals (including normal sized
well shaped ice crystals) and their nuclei measured by electron microscopy,
regardless of temperature, location and other conditions.
3. Sizes and Composition of Ice-Fog Nuclei and Electric Conductivity
of Melted Ice-Fog Crystals:
i
The sizes of ice-fog nuclei are quite small in comparison with those
of normal fog in cities of the temperate region, but similar to mountain
fog or cloud nuclei. Figure 17 shows the size distribution of ice-fog
nuclei. These sizes are almost in agreement with Kumai's result. However,
-------
CO
100 r
CO
a: 80
o
Ł 60
i
LU
a
u. 40
O
CO
Ł 20
CD
D
z n
-
-
-
-
I
mm
mm
_„
••i
B^H
mm
mm
mm
mm
-Hi
Tm-rh-j-
5 10 15 20 25 30 35 40
DIAMETER (JJL) (ASSUMED ALL CRYSTALS WERE SPHERICAL)
Fig. 16. Size distribution of ice-fog crystals examined under the electron
microscope.
-------
160
140
120
^100
o
80
ffi 60
m
40
20
0.5 1.0
1.5 2.0 2.5
DIAMETER (p.)
3.0 3.5 4.0
Fig. 17. Size distribution of ice-fog nuclei examined under the electron
microscope.
-------
generally speaking, it is difficult to determine whether a particle is a
nucleus of an ice-fog crystal or a contamination particle in or on the
collodion and formvar films. Even though we tried to make the films on
sheet mesh (grid) as clean as possible, sometimes many minute particles
could be found outside and inside the replicas of crystals. Based on the
author's experience of electron-microscope study of fog and cloud nuclei,
the nuclei of ice-fog crystals were distinguished from the contamination
particles by aid of replication and shape of particles. However, since
identification was difficult in some cases, we measured the most probable
and the largest particle inside the replica. After improving our technique
we found many ice-fog crystals containing no particles inside the crystal
replica. The crystals with no particle inside of them show that the cry-
stals were frozen spontaneously from supercooled water droplets which had
homogeneously condensed from water vapor or sublimed onto small fragments
of ice particles, although the last mechanism is not likely in the case
of ice fog. If any small particle was found in the crystal we assumed it
was a nucleus. This means that we might have underestimated the possibility
of spontaneous nucleation in the formation of ice-fog crystals. However
any particles outside of the replicas of ice-fog crystals were disregarded
as contamination or as dusts which co-existed with ice fog in the air.
Generally speaking, taking pictures of nuclei of ice-fog crystals with the
electron microscope without replicas and referring to optical micrographs
as Kumai (1964) did would give errors in the original locations of ice-fog
crystals by as much as 5y. This is dangerous, especially for the smallest
sized crystals.
44
-------
As can be seen in Table 6, 71 out of 713 crystals inspected by the
electron microscope did not have any nucleus in the crystal replicas.
Most of these were found in ice fog at Chena and Manley Hot Springs at
temperatures lower than about -40C. An example of ice-fog crystals is
shown in Fig. 18. At the IAP site which is located near the open water
area along the Chena River, we found no nuclei in about 12 percent of 110
ice-fog crystals we collected when the temperature was below -39C. At
other places such as on the University of Alaska campus we could sometimes
find only very small nuclei or contaminations in the crystal replicas. At
the center of city only 1.7 percent of 236 ice-fog crystals had no nuclei,
and a few ice-fog crystals had many particles inside of the replicas and
few or none outside. An example is shown in Fig. 19. These crystals were
probably frozen from dirty droplets formed directly from some kind of
exhaust such as car exhaust. On the other hand many ice-fog crystals even
in the center of the city did not have large dust particles. These ice-fog
crystals must have formed initially from the process of water vapor
condensation under conditions of high supersaturation which are available
from car, power and heating plant exhausts and from steam coming from open
water.
Although we tried to identifyithe composition of the nuclei by use
of the electron microscope and electron diffraction, we could not determine
the composition very well, especially for the smallest size nuclei. Some
nuclei were presumed to be only carbon black (Fig. 20) resulting from
combustion. Using a morphological determination which was essentially the
45
-------
TABLE 6
Percentage of composition of ice fog nuclei
microdiffraction. At the MUS site 1C means
nuclei.
Site
MUS site *Ł
lr
ZAP and slough
MUS power plant
U. of Alaska
BLM (along
Highway)
Int. airport
Manley H. S.
Chena H. S.
Total
Total
10
236
120
52
84
32
59
32
88
713
Combustion
by-product%
5 50
222 94
96 80
50 96
73 87
27 85
54 92
22 70
50 57
599 84
determined by electron microscope and electron
ice crystal nuclei and IF means ice fog crystal
Soil Particle
%
5 50
6 3
5 4
2 4
3 4
2 6
1 2
0 0
0 0
24 3
No nuclei %
0 0
4 2
13 11
0 0
2 2
2 6
3 5
9 28
38 43
71 10
Under termined %
0 0
4 2
6 5
0 0
6 7
1 3
1 2
1 3
0 0
19 3
-------
Fig. 18. Ice-fog crystal without nucleus, col-
lected at Chena H. S. on December 31, 1968.
Temperature was -45C.
Fig. 20. Ice-i'og crystal with carbon black as
a nucleus. The crystal was collected near the
IAP site on December 8, 1968. The white part
at upper right of picture seems to be a kind of
spit-ule formed from the crystal.
Fig. 19. Ice-fog crystal with many dust parti-
cles only inside of the crystal. The crystal
was sampled at the MUS site, on January 2,
1969. Temperature was -45C.
Fig. 21. Spherical ice-fog crystal with very
small nucleus off center. The crystal was
collected near the IAP site on December '2 1 .
1965. Temperature was -36. 4C.
-------
same as Yamamoto and Ohtake's (1953), we found that most (more than 90
percent) of the Ice-fog nuclei were classified as combustion by-products.
Kumai (1964) found 74 percent and 90 percent of the ice-fog nuclei were
combustion by-products for the temperatures of -37C and -41C, respectively.
In addition he reported 21 percent and 10 percent of the nuclei were clay
minerals and he found no ice-fog crystals without nuclei at the temperatures
of -37C and -41C. On the other hand we found only a few soil particles
as ice-fog nuclei, and more than 80 percent and 10 percent of the total
number of ice-fog crystals had combustion by-product nuclei and no nucleus,
respectively. A possible reason why Kumai did not find any ice-fog crystals
without nuclei may come from his direct sampling on collodion film without
replication. He might have missed the exact original positions of the
ice-fog crystals on the collodion film under the electron microscope.
With our electron-microscope studies, there was no evidence of many
chance dust particles or solid phase "air pollution" (other than ice fog)
deposited on the surfaces of the ice-fog crystals. It is quite possible
that the particles considered as contamination on the formvar films were
the dust particles which coexisted with ice-fog crystals and were pre-
cipitated by small air turbulence onto the film and in rare cases on a
replica of an ice crystal. If this is correct, then the particles should
be randomly distributed on the film.
If we make some assumptions about the size and composition of the
nuclei and the size of the ice-fog crystals, we can derive the concentration
and thus the electrical conductance of the melted ice-fog crystals. If we
48
-------
assume the composition of the nuclei to be KC1 or H-SO,, normalities of
these solutions should be 4.16 x 10~*N (= 3.1 x 10~2 gm *-1) and
5.83 x 10 N (= 2.9 x 10~2 gm 2-"1) for KCl and H2SO,, respectively,
assuming the average diameters of ice-fog crystals are 8y and of ice-fog
nuclei are 0.2y, which were the estimated peaks of the size distributions
of Figs. 16 and 17. Using these values we get the specific electrical
conductance of melted ice-fog crystals as 54 ymho cm and 225 ymho cm
for all the nuclei assumed to be KCl and H2SO,, respectively. These values
are of the same order of magnitudes as Benson's (1965) results from melted
and filtered snow pack in the city. He reported the electric conductances
were about 100 ymho cm in the city and 10 ymho cm in the outlying
area; he also reported that a lot of water insoluble particulate matter
was concentrated at the center of the city. In our data, however, all
nuclei were assumed to be water soluble KCl or H-SO, and their apparent
diameters were measured as an equivalent circular area, after which we
calculated the volume of the particles as spheres. Also the aqueous
solution of H_SO, was selected as the possible chemical material which
gives the highest value of equivalent conductance of an aqueous solution.
From this point of view, even though the present result agreed with Benson's
(1965) observation, all the above mentioned values must have been overestimated
in the present estimation due to the many assumptions in our reduction and the
many water soluble and insoluble dust particles precipitated separately
from ice-fog crystals in the case of Benson's. On the other hand, our
analysis considered only visible nuclei, neglecting the possibility of
49
-------
ions adsorbed on the crystals. For crystals on the order of 8y, absorption
effects would be negligible.
4. Position of Nuclei in Ice-Fog Crystals:
The positions of the nuclei of ice-fog crystals were examined by
means of the replica technique and the Specimen Position Indicator which
was previously described. Many nuclei of ice-fog crystals smaller than
about lOy diameter were found not in the center but rather away from the
center of the crystals, especially in ice-fog crystals which had been
collected when the air temperature was lower than -35C. Figures 21 through
24b show some examples. Of course ice-fog crystals with no nucleus are
excluded. In contrast the nuclei of ice crystals larger than lOy diameter
(which normally have well developed crystal faces) were at the exact
centers of the ice crystals. Examples are shown in Figs. 25 and 26. This
was also confirmed by direct collection with the aid of the Specimen
Position Indicator. The probable explanations for these positions of
nuclei may be as follows: The smaller ice-fog crystals must result from
the freezing of supercooled water droplets. The supercooled water droplets
are formed on condensation nuclei through water vapor condensing onto the
nuclei. When the condensation nucleus grew into a water droplet, the
nucleus could move by Brownian motion anywhere inside of the water droplet,
because the water droplet is in the liquid phase. At the moment the water
droplet freezes under temperatures lower than -30C there is no reason for
the center to be a preferential position for the nucleus. This explanation
50
-------
Fig. 22. Ice-fog crystal and its
nucleus sampled at the Fairbanks
International Airport on Jan. 1,
1966. The nucleus is presumed to
be a combustion by-product. The
temperature was -43. 2C.
Fig. 23. A spherical ice-fog crystal
with off center nucleus. The nucleus
is presumed to be combustion by-
product. The sample was taken near
the Noyes slough in winter of Jan. 1965.
The temperature is unknown.
Fig. 24a. Nuclei of an ice-fog crystal.
The nuclei were presumed to be com-
bustion by-products, which may have
been in the gas phase in temperate
weather. Original outline of ice crys-
tal also illustrated.
Fig. 24b. Enlargement of the nuclei
of Fig. 24a. The crystal was col-
lected directly onto collodion film at
the MUS site on Feb. 21, 1966. The
temperature was -32. 9C.
-------
Fig. 25. Ice-fog crystal with nucleus at
center. The nucleus was also presumed
to be soot. The crystal was sampled at
the MUS site on December 23, 1965.
Fig. 26. Ice-fog crystal with nucleus ;
the center. The nucleus was presumed
to be combustion by-product. The sam»<
was taken near the University power plH
on Jan. 6, 1966. The temperature was
-39. 8C and visibility was 200 m.
Fig. 27. Ice-fog crystal with spicule.
The crystal was sampled on January
16, 1968 at the MUS site at a tempera-
ture of -38. 2C.
Fig. 28. Ice-fog crystal (replica) with
boundaries or creases. The sampling
was made at Chena H. S. on 31 Decembe:
1968 at a temperature of -45C.
-------
strongly supports the mechanism of formation of ice-fog crystals as being
the freezing of water droplets. So far most ice-fog investigators have
suggested the mechanism of ice-fog crystal formation to be the freezing
of water droplets, but they did not have any strong observational evi-
dence.
Another possible explanation for the off-center nuclei observed in
many crystals would be general contamination of formvar films with many
homogeneously nucleated crystals. However, the density of contamination
on the portion of the formvar films not covered by crystals is too low to
account for more than a small fraction of the observed off center nuclei.
Larger well formed crystals which have nuclei at their centers may have
grown by sublimation onto very small ice-fog type crystals which probably
passed through a liquid phase rather than directly sublimating onto nuclei.
If the droplet was small enough, the displacement of the nucleus from the
center of the final large crystal would be undetectable. This concept
represents a revision of opinion since Ohtake (1967), in which it was
assumed that centered nuclei indicated initial crystal formation by sub-
limation.
5. Spicules on Ice Fog Crystals:
Many nearly spherical or columnar ice-fog crystals appeared to have
a sort of projection when viewed under the electron microscope (Henmi,
1969). Because there was some possibility that the projections were formed
only on replicas of ice-fog crystals, it was confirmed under the optical
53
-------
microscope that many ice-fog crystals suspended in silicone oil had such
projections or spicules. Figure 27 is an example of a crystal with a
spicule which was probably produced by the Bally-Dorsey effect (Dorsey
1948,Blanchard 1951). Concerning the spicules, Dorsey stated as follows:
"As freezing proceeds, the pressure of the enclosed water rapidly increases,
the temperature changing but little. The pressure may rupture the surface
ice at some weak spot. A jet of very slightly supercooled water then issues
through the break, its surface freezes promptly, forming a tube which grows
at its tip, and through which water continues to flow until the pressure
is sufficiently relieved or the tube has become blocked with ice." In
the case of small droplets, say lOy, at -40C temperature the surface
tension is strong and the blocking of the tube rapid so that the spicules
may not grow as large as for the case of larger drops. All the fog
particles found at large enough distances from the water source were solid
ice (see chapter 8), which effectively rules out the possibility that the
spicules were formed on unfrozen droplets hitting the slides.
Spicules thus appear to be an excellent indicator that particles
carrying them have frozen from a supercooled liquid phase. The crystals
shown in Figs. 20 and 27 were almost certainly formed from supercooled
droplets. On the other hand, the crystals shown in Figs. 25 and 26 do
not have spicules, and may have grown by sublimation after initial freezing.
This phenomenon also supports the idea that the ice-fog crystals originate
from freezing of supercooled water droplets. The size and shape of ice-fog
/
crystals near open water are essentially the same as those of water droplets
-------
very near the open water, although near open water in the city we can find
other sizes and shapes of crystals which have drifted from beyond the open
water.
6. Sintering of Ice-Fog Crystals and Structure of Ice-Fog Crystals:
Some electron photomicrographs of ice-fog crystal replications had
a line or several lines as can be seen in Figs. 19 and 28. The crystal
of Fig. 28 was collected in replica at Chena Hot Springs, and has many
lines on the picture. These lines are likely to be boundary lines of ice-
crystal grains, because the ends of the lines seem to separate the crystal
into several parts. Another crystal which was also taken at Chena Hot
Springs and has no nucleus is shown in Fig. 29. This crystal has a line
in the middle which appears to divide the crystal into two coagulated
parts. Under an optical microscope we could not find many of these lines
because of the poor resolving power compared with that of an electron
microscope. The explanation of this appearance has not been clarified.
Kumai (1964) reported 65.5% of the ice-fog crystals sampled at -39C
in Fairbanks were combined crystals (formed by sintering of two or more
spherical crystals) and 8.5% of those were columnar crystals with a bound-
ary (formed by sintering of two spherical ice crystals). Hobbs (1965)
supported Kumai's observation by theoretical considerations. However the
photomicrograph in Fig. 7a of Kumai's paper seems to show that the slide
glass had been exposed to ice fog for too long a time and probably many
ice-fog crystals stuck together on the slide rather than in the air. In
fact in Kumai's Fig. 7b which was taken at the temperature of -41C such a
55
-------
Fig. 29. Ice-fog crystal at Chena H. S. The crystal has a line in the
middle and was found on the same specimen mesh as the
crystal shown in Fig. 28.
o o o
Fig. 30. Optical microscope picture of ice-fog crystals at Chena H. S.
Temperature was -45C.
-------
high percentage of "aggregated crystals" does not appear. This must
result from the collection of an appropriate number of crystals by
precipitation on the glass slide. According to the present observations
(see next chapter), irregular shaped crystals were only 6% and columnar
crystals with a boundary on each crystal were less than 8% for the
temperature range of -38 to -40C. One might attribute the scarcity of
the aggregated crystals of Kumai's Fig. 7b to the lack of a water film
between ice particles to be bounded at the "critical temperature -41C
as spontaneous freezing".
Kumai considered that the combination of two small spherical ice-
fog crystals through the sintering process resulted in a column crystal
with a boundary, and if three crystals were sintered, the aggregation
would produce a three pointed bullet crystal. Such sintering of crystals
will occur even at -40C as Hobbs stated if the crystals collide. Hobbs
estimated that the time needed for sintering to occur between two ice-fog
crystals of 5y radius each at -40C must be 760 seconds. Furthermore,
according to him, this time of 760 seconds is easily expected in ice fog
in which crystals fall in the 10 m thickness (height) of usual ice fog.
Even though the available travel time of ice-fog crystals is much more
than 760 seconds because the ice fog is normally 50 m thick and crystals
are usually drifting almost horizontally, in order to sinter two ice-fog
crystals, they should contact or almost contact each other for as long as
760 seconds. This is almost impossible in the actual ice fog.
57
-------
Thus now we have arrived at the point of checking the possibility
of collision of two ice-fog crystals. Since the average falling speed
of ice-fog crystals is less than the turbulent velocity of air in ice fog
as was described in the previous chapter, their relative motion should be
controlled by Brownian random motion of each ice-fog crystal rather than
aerodynamical motion due to differences of falling speeds. Fletcher (1966)
summarized the theory for coagulation of aerosols, using the development
of a basic equation by Smoluchowski for liquid suspension and their appli-
cation to aerosols by Whytlaw-Gray and Patterson, and to give an idea of
numerical magnitudes gave an equation for reasonably homogeneous aerosols
as follows:
Here r is the radius of an average particle in cm, n is the particle con-
_3
centration in particles cm and t is time in hours. From this equation if
we assume a particle diameter of lOy and initial concentration of 1500
_3
particles cm , we find that it takes about 220 hours for the concentration
_3
to decrease to 1000 particles cm (after coagulation, 50% of the particles
are two-particle aggregates). 220 hours is far beyond the lifetime of a fog
particle, or even of most ice fogs. Also Fletcher (1966) remarked on the
coalescence of water droplets:
-3
"In a typical cloud there may be perhaps 300 droplets cm , all
roughly lOy in diameter. Equation (4.12) (same as the above
equation) then states that the collision efficiency is only
0.1 cnr3hr~-1- or 30 nT^sec"*. This represents a completely
58
-------
negligible proportion of the cloud droplets during the life of an
ordinary cloud, so that second collisions with these larger droplets
virtually never occur. Collisions due to Brownian motion can thus
be neglected as a mechanism for the production of appreciable numbers
of larger droplets. With Brownian motion neglected and the effect of
gravity considered, the problem now becomes an aerodynamic one".
The microscope picture (Fig. 30) of ice-fog crystals which was taken at
Chena Hot Springs at -45C, not only fails to show many aggregated crystals,
but also has no columns with boundaries. Thus we arrive at the conclusion
that the sintering of ice crystals may be neglected in the case of ice fog.
The columnar crystal with a boundary at the middle is considered to be a
twin crystal rather than a crystal formed through the sintering process.
Even though we had a narrow monotonic peak in the size distribution
of ice-fog crystals at Chena Hot Springs, which indicates the infrequency
of aggregation in both the liquid and solid states of water, we found
lines similar to grain boundaries in the smallest sized ice crystals, as
can be seen in Figs. 28 and 29. Considering the points discussed above,
it is suggested that these grain boundaries may be attributed to abnormal
thermal distribution in the water droplets or ice crystals at the time of
their freezing. This concept would also cover the mechanism of formation
of twin or poly-»crystalline crystals. An attempt was made to look at such
crystals under a polarizing microscope. All columnar crystals showed mono-
orientation. However, a few irregular crystals showed a brilliant portion
and one or more dark portions with different angles of rotation of the
microscope stage. Figure 31 shows a sketch of such ice-fog crystals. The
three different parts were brilliant alternately with rotation of the stage.
This crystal was presumed to be composed of three differently oriented
59
-------
ABOUT 40/x
Fig. 31. Sketch of two ice-fog crystals under a polarizing
microscope. The crystals were collected at the
MUS site at a temperature of -37C.
Fig. 32. Nuclei which were not the ice-
fog nuclei in ice fog.
Fig. 33. Optical microscope picture of
the types of ice-fog crystals taken at
the MUS site. Temperature was -33. 5C.
HP. : Hexagonal Plates, Sp. : Sphericals,
P. C. : Plain Columns, S. C. : Skeleton
Columns and Irr. : Irregular Crystals.
-------
crystal grains. The approximate fraction of these crystals was only a few
percent of the total ice-fog crystals at -37C. On two occasions we found
that the bright portion continuously moved from one side to another as
the stage was rotated. Those must have been wedge-shaped column crystals.
Unfortunately, there was no chance to use a good polarizing microscope,
especially for taking a picture at high magnification. Another possible
explanation is that the formvar film may have developed wrinkles during
the replication process, even though Fig. 29 does not seem to support
this idea. Further study is needed. The question might be solved by use
of a scanning electron microscope.
d. Inactive Nuclei in Ice Fog:
ft ^
As was previously mentioned, we have about 10 particles cm of
total condensation nuclei in the atmosphere of Fairbanks city. However,
the concentration of ice-fog crystals during ice fog was about 200 crystals
_3
cm . The role of the inactivated nuclei is of interest. For this reason,
the inactive nuclei during ice fog in the air filtered by a Millipore fil-
ter, pore size 14y, have been taken on specimen grids for examination by
electron microscope. The use of the Millipore filter to restrict ice-fog
crystals from coming into the specimen grid also removes some of the nuclei
as mentioned before. Although we have only three electron micrographs for
them, the inactivated nuclei were presumed, from their shape, to be com-
bustion by-products from car exhaust or other hygroscopic materials. And
such nuclei seemed to be essentially of the same composition as ice-fog
nuclei. See Fig. 32 as an example.
61
-------
7. OPTICAL MICROSCOPE STUDIES OF ICE-FOG CRYSTALS
A. The Effect of Temperature and Humidity on Ice-Fog Crystals
1. Sampling Method for Ice-Fog Crystals:
In order to study the size distributions, the mean diameters, the
precipitation rates and the concentrations of ice-fog crystals, the crystals'
were collected on microscope slides covered by a thin film of silicone oil.
A box of size 10 cm x 10 cm x 10 cm was used to shelter the slides from
surface winds. In most cases, exposure times were 20 minutes, after which
the ice fog crystals collected on the slides were immediately photomicro-
graphed. This was a use of the precipitation method which has many dis-
advantages compared with the methods previously mentioned.
According to our pre-observations, there was apparently a temperature
dependence of the solid water content of ice fog, i.e. we had no ice fog,
slight and heavy ice fog at the temperatures of -20, -30 and -40C,
respectively (refer to later chapter). Thus, it is necessary to know the
temperature dependence of concentrations, precipitation rates and solid
water content of ice fog. By the precipitation method, it is impossible
to obtain the absolute values as indicated previously, but it may be
possible, if we use a large amount of data, to obtain approximate average
values.
Ambient temperatures were measured by means of a bimetallic thermo-
meter whose accuracy was better than 40.2C. Temperatures and most
humidities were observed simultaneously when ice-fog crystals were sampled,
so that the effect of ambient humidity on ice-fog crystals could be studied.
62
-------
2. Method of Analysis:
a. Determination of the Diameters:
The diameters of ice-fog particles were measured from the photo-
micrographs enlarged to a magnification of 300. The Carl Zeiss TGZ3
Particle Size Analyzer was used to determine the diameter of each crystal.
To obtain mean diameters for each sampling time, 500-2000 ice-fog crystals
were measured.
b. Classification of the Shapes of Ice-Fog Crystals:
Thuman and Robinson (1954c) classified ice-fog crystals into the
following three types: Hexagonal plates, prismatic columns, and droxtals
(equant solid particles with rudimentary crystal faces). Kumai (1964)
studied the size distributions of two types of ice-fog crystals, hexagonal
plate and spherical.
It is, however, necessary to classify ice-fog crystals more precisely
in order to study their growth process. Therefore, in the present work,
ice-fog crystals are classified into the following five types:
1. Hexagonal plates
2. Plain columns
3. Skeleton columns (columns with inner designs
or a boundary)
4. Sphericals
5. Irregulars
"Irregular" means that the particles cannot be classified into any
of the four preceding types. In Fig. 33, a photograph of typical re-
presentatives of each type is shown.
63
-------
3. Results and Discussion:
a. Mean Diameters of Ice-Fog Crystals:
Sixty-five samples were taken in the temperature range -30 to -41C
under ice-fog conditions. The relationships between the mean diameters
of different types of ice-fog particles and temperatures are shown in Figs.
34 to 39. From these figures, the following facts may be inferred:
1. Although data are scattered, the mean diameter of ice-fog crystals
is obviously a linearly increasing function of temperature.
2. The dependence on temperature of sphericals, however, is much
less than that of other types of crystals.
3. The mean diameters of skeleton columns are larger than those of
plain columns.
4. Hexagonal plates show the largest mean diameters.
5. Above the temperature of -35C, the mean diameters of each type
are more scattered than below -35C.
It is probable that the scattering of the mean diameters is mainly
due to the variation of wind direction at the sampling time, because there
are many sources of ice-fog crystals.
According to Thuman and Robinson (1954a), the mean diameters of ice-
fog particles are statistically logarithmic functions of temperature. The
mean diameters obtained here are slightly smaller than those obtained by
them. In addition, their results show that generally the largest crystals
are columns and the second largest are hexagonal plates. On the contrary,
our results show that the largest are hexagonal plates, and the second
largest are skeleton columns. For purposes of comparison, their results
are shown in Figs. 35 and 38.
64
-------
o:
LL)
I-
LU
40 r
35
30
25
20
UJ
10
o o
o o
TOTAL
-40 -35
TEMPERATURE (°C)
-30
Fig. 34:
Relationship between mean diameters of total ice-fog crystals
and temperatures.
-------
40
35
30
^ ^
3.
o:
I-
LJ
5
25
20
<
LU
15
10
THUMAN
ROBINSON
00
HEXAGONAL PLATES
-40 -35 -30
TEMPERATURE (°C)
Fig. 35. Relationship between mean diameters of hexagonal plates
and temperatures.
-------
-.5
tr
UJ
i-
LU
o
z
<
LU
10
THUMAN SPHERICALS
ROBINSON
o
I I
-40 -35 -30
TEMPERATURE (°C)
Fig. 36. Relationship between mean diameters of sphericals and
temperatures.
-------
30
25
20
tr
UJ
i-
LU
15
10
SKELETON /
COLUMNS x' o
X
o o
PLAIN COLUMNS
-40 -35 -30
TEMPERATURE (°C)
Fig. 37. Relationship between mean diameters of plain columns and
temperatures.
-------
SKELETON
COLUMNS
o/o x''PLAIN COLUMNS
-40 -35 -30
TEMPERATURE (°C)
Fig. 38. Relationship between mean diameters of skeleton columns
and temperatures.
-------
tr
UJ
UJ
45 r
40
35
30
<
yj
20
10
o o
IRREGULAR SHAPE
-40 -35 -30
TEMPERATURE (°C)
Fig. 39. Relationship between mean diameters of irregular shaped
crystals and temperatures.
-------
The differences between their results and ours are probably due to
the following reasons: 1) Different location for sampling, 2) different
classification of types, and 3) different scaling methods. Thuman and
Robinson defined the diameter as the mean length of a line which approxi-
mately bisected the area of the profile of an ice-fog crystal. The bi-
secting line is taken parallel to a fixed direction, irrespective of the
orientation of each crystal. However, in our case, the diameter is defined
as that of a circle whose area is equal to that of the profile of an ice-
fog crystal.
In an experiment using a diffusion cloud chamber, Kobayashi (1956)
observed the following successive changes in shape of the precipitating
ice crystals after silver iodide seeding:
Droxtals and hexagonal columns (or plain columns)
^ Hexagonal simple plates
^ Hexagonal simple plates with skeleton structure
^ Sector forms and droxtals.
The interesting point is that hexagonal simple plates with skeleton
structure are formed after hexagonal simple plates. This seems to suggest
that the skeleton columns are formed by the growth of plain columns, which
explains the fact that the mean diameters of skeleton columns are larger
than those of plain columns. Furthermore, from the change of ice crystal
shapes observed by Kobayashi and the fact that the mean diameters of
sphericals are almost independent of temperature, it is inferred that the
sphericals represent the initial stage of growth.
71
-------
b. Size Distribution and Percentage Distribution of Types of
Ice-Fog Crystals:
Figures 40, 41 and 42 show the mean size distributions of ice—fog
particles in the temperature ranges -31.0 to -32.9C, -35.0 to -36.9C
and -39.0 to -41.OC, respectively. Each figure is obtained by averaging
7 to 10 different sets of data. From the figures it is obvious that the
lower the temperature becomes, the narrower is the breadth of distribution
of crystal size.
From the percentage distribution of five ice-fog crystal types, which
is shown in Fig. 43, it is evident that the percentage of sphericals becomes
suddenly predominant below -37 or -38C. Such spherical crystals must be
formed near the observation site through the freezing of supercooled water
droplets which then undergo very little further growth. The temperature
of -37 or -38C may be considered as a threshold temperature of the small
nuclei being active as freezing nuclei in the supercooled water droplets
or that of the spontaneous nucleation (homogeneous freezing) of 10v diameter
water droplets which do not have any nucleus. Furthermore, the ratio of
the percentage of skeleton columns to that of plain columns decreases with
decreasing temperature, which seems to imply that the probability of the
growth from plain column to skeleton column decreases with decreasing
temperature.
c. Numbers of Precipitated Ice-Fog Crystals:
Using the data obtained by the precipitation method, the number of
precipitated ice-fog crystals N is plotted against temperature in Fig.
rl
44. It can be seen that although the data are scattered, there is a
definite tendency for Nu to increase exponentially with decreasing
72
-------
-31.0 32.9° C
TOTALS
PLAIN COLUMNS
SKELETON COLUMNS
HEXAGONAL PLATES
SPHERICALS
IRREGULARS
I0r
UJ
o
o:
UJ
o_
TOTAL
15 25 35
DIAMETER (MICRON)
Fig. 40. Mean size distributions of various shaped ice-fog crystals
for temperature range -31. 0 to -32. 9C.
-------
-35.0 36.9° C
TOTAL
TOTALS
PLAIN COLUMNS
SKELETON COLUMNS
HEXAGONAL PLATES
SPHERICALS
IRREGULARS
15 25 35
DIAMETER (MICRON)
45
Fig. 41. Mean size distributions of ice-fog crystals for temperature
range -35. 0 to -36. 9C.
-------
20
15
o 10
o:
LU
QL
TOTAL
-39.0 41.0 °C
TOTALS
PLAIN COLUMNS
SKELETON COLUMS
HEXAGONAL PLATES
SPHERICALS
IRREGULARS
5 15 25 35
DIAMETER (MICRON)
Fig. 42. Mean size distributions of ice-fog crystals in various shapes
for temperature range of -39. 0 to -41. OC.
-------
FLEXIBLE
E3 IRREGULAR SHAPES
(^ SPHERICALS
• HEXAGONAL PLATES
SKELETON COLUMNS
PLAIN COLUMNS
100
-40 -35
TEMPERATURE (°C)
-30
Fig. 43. Percentage distribution of the shapes of ice-fog crystals.
-------
5000 c
o o
-40
-35
TEMPERATURE (°C)
-30
Fig. 44. Numbers of ice-fog crystals precipitated versus ambient
air temperatures.
-------
temperature. The change in N_ with temperature is given by the following
.equation derived empirically from Fig. 44.
M^ - 4.7 x 10"2 exp (-5.4 x 1(T2T)
where T = temperature in degrees Celsius.
This increase in the number of precipitated crystals with decreasing
temperature might be expected for two reasons: 1) the lower saturation
vapor pressure at lower temperatures would lead to the production of more
droplets for the same vapor input, and 2) the probability of activation
of air pollutants as freezing nuclei, as well as the number of natural
freezing nuclei, increases with decreasing temperature (Fletcher, 1962).
According to prior work (Benson, 1965), data obtained by the precipitation
method gave smaller values than are correct, due to upward winds. There-
fore, in reality the gradient of the line may be steeper than that in
Fig. 44, because there are more small particles at low temperature than
at higher temperature. However, for temperatures lower than about -40C
the numbers of precipitated crystals tended to be constant in downtown
Fairbanks. This might be due to a drop in the number of automobiles
running. In addition, if spontaneous freezing takes place around -40C
reason (2) above is no longer valid below -40C and a change in the form
of dependence of N« on T would be expected. The tendency for the curve
in Fig. 44 to flatten for temperatures lower than -40C might be due to
spontaneous freezing.
d. Precipitation Rate and Solid Water Content of Ice-Fog Crystals:
Next, we calculate the precipitation rate and the solid water
content of ice fog, assuming the lines drawn in Figs. 34 and 44 give the
78
-------
average values of the mean diameter and precipitated number of ice-fog
particles, respectively, as functions of temperature.
From Stokes* law, the terminal velocity v of a spherical ice
particle is given by
VT
2 pi - pa
where
p. = density of ice
p = density of air
TI = viscosity of air
g = acceleration of gravity
r = radius of ice-fog particle
—2 -1
The precipitation rate R(g cm min ) of ice fog crystals is given by
when N.. is the number of ice crystals precipitated on a unit area during
i _3
unit time. The solid water content (g cm ) of ice fog is given by
Using these equations, the precipitation rate R and the solid water content
W are obtained and listed in Table 7 and drawn in Fig. 45. The precipitation
79
-------
TABLE 7
Mean Values of Precipitation Rates and Solid Water Content of Ice Fog (by precipitation method)
Temperature
(°C)
-30 (in city)
-32 (in city)
-34 (In city)
-36 (in city)
-38 (in city)
-40 (in city)
(-47) (In city)
Kumai -39 (in city)
Kumai -41 (airport)
Benson <-30 (in city)
Benson <-35 (In city)
Benson (outlying area)
Crystals
Precipitated
-2 -1
NH (cm min )
240
425
750
1350
2300
4200
(3889)
2302
832
no observation
no observation
no observation
Mean Diameter of
Ice-Fog Crystals
D <„>
32.7
28.0
23.3
18.5
13.9
9.1
-
-
—
-
Terminal Velocity for
Mean Size of Ice-Fog
Crystals _
VT (cm sec )
3.47
2.55
1.78
1.12
0.62
0.27
-
-
+
+
?
Precipitation
Rate-2 -1
R (jig cm min )
4.01
4.43
4.26
4.08
2.94
1.51
0.691
0.0287
2.18 (mean)
2.94 (mean)-^
0.02
Solid Water
Content
W(g m"3)
0.019
0.029
0.043
0.061
0.078
0.094
(0.055)
0.07
0.02
0.21
0.07
00
o
-------
,-JOr
PRECIPITATION RATE
SOLID WATER
CONTENT
-40 -35
TEMPERATURE (°C)
-30
Fig. 45. Solid water contents and precipitation rates of ice fog.
-------
—2 —1
rates were of the order of 1 yg cm min . The solid water contents were
_0
of the order of 0.01 g m and Kumai (1964) made similar calculations of
the solid water content, using size distributions instead of the mean
diameters used here. He reported that the solid water content is 0.07
-3 -3
g m and 0.02 g m at -39C and -41C, respectively. It must, however,
be noted that Kumai*s results were obtained by single sampling and do
not represent average values of many data at these temperatures. Benson
_o
(1965) estimated 0.21 g m for the core area of Fairbanks. Since this
value was not derived from observation, it seems to be overestimated.
The liquid water content in clouds has been extensively studied. Fletcher
(1962) summarized the measurements made by many authors. According to his
_3
table, the liquid water content in clouds is of the order of 0.10 g m ,
which is an order of magnitude larger than that of ice fog.
Precipitation rates were also studied by Kumai (1964) and Benson
(1965). Kumai obtained 0.691 yg cm min , and 0.0287 yg cm min at
—39C and -41C, respectively. These values are one to two orders of mag-
nitude smaller than the results obtained here. Benson directly measured
2
precipitation rates, using aim plastic sheet as a collector. The
-2 -1
precipitation rates he obtained are about 2 to 3 yg cm min , which is
the same order as obtained here. The reasons that Kumai's results are
smaller than those of Benson and those obtained here may be due to the
fact that the concentration of ice fog was small when he sampled, and also
due to the upward winds in the ice-fog layer or variation in sample
location, as described by Benson.
82
-------
e. The Effect of Ambient Humidity on Ice-Fog Crystals:
So far, we have described the effect of temperature on ice-fog
crystals. Now let us consider the effect of ambient humidity on ice
fog crystals. As is well known, the theoretical equation for the growth
rate of an ice crystal in water vapor is given as follows (Byers, 1965):
~ = 4irCD (p -po)
where
M is the mass of the ice crystal
C is the capacity (dependent on crystal shape)
D is the diffusivity of water vapor in air
p is the vapor density of ambient air
p is the vapor density at the surface of the droplet
Therefore, there may be some relationship between the mean diameters
of ice-fog crystals and ambient humidities.
In order to study this ,v Henmi (1969) plotted the mean diameters of
different types of ice-fog crystals--against supersaturation over ice
for the temperature ranges -30.0 to -32,OC, -32.1 to -35.OC, -35.1 to -37.OC,
-37.1 to -39.OC, and -39.1 to -41.OC. In these figures, he assumed the
effect of temperature on ice-fog crystal growth to be the same in each
temperature range. The precipitation rate of ice-fog crystals was also
plotted against ambient supersaturation over ice and the resulting plot
is given here as Fig. 46.
These figures seem to indicate that the mean diameters of ice-fog
crystals were independent of ambient humidity, and that there was no
83
-------
ouuu
^
| 1000
CM*
5
o 500
a!
LU
h-
o:
0 100
E
o
LU
a:
Q_
o o
i ° °
° 0 °
0 ^ 0 ° ° ° °
c °°* * °
0 ° ° rf>
o
o
o
o
o
0 0
0.0 O.I 0.2 0.3 0.4 0.5
SUPERSATURATION OVER ICE
Fig. 46. Numbers of ice-fog crystals precipitated versus super-
saturations over ice.
-------
relationship between the precipitated crystals and ambient humidity.
Therefore, it is probable that ice-fog crystals are almost completely
developed around the sources, and that after dispersing away from the
sources, they are only drifting without significant growth or evaporation.
Had we observed the humidities and the mean diameters near water vapor
sources, some relationships might have been found.
f. Ice-Fog Crystals from an Unpolluted Area tfith a Moisture Source:
We normally see dense ice fog only in inhabited areas such as the
Fairbanks vicinity. This is the reason why ice fog is considered to be
a form of "air pollution" and 'the importance of pollution in forming ice
fog is emphasized. However, around the hot springs at Chena and Manley
Hot Springs or along open water in a river (a rapidly running stretch or
a small portion of open water made by overflow above thick ice), dense ice
-3
fog occurs even in very clean air with only 50 to 500 p cm condensation
nuclei. The size distribution of such ice fogs formed in clear air had
a narrow peak at a diameter of about lOy. All particles were spherical
or nearly spherical (Figs. 30 and 47). The crystals with spicules did
not seem to be aggregated from similar sized crystals or droplets but to
be formed by expansion from mother crystals. A similar size distribution
i
of crystals also occurs at the IAP (open water) site in Fairbanks. Normally
at the MUS site in the city the size distribution has three peaks which
correspond to ice-fog crystals originating from (1) automobile exhaust,
(2) open water, and (3) heating and power plant exhausts.
85
-------
Fig. 47. Ice-fog crystals at Chena H. S. on December 31, 1968.
Fig. 49. A polyhedral crystal looking from the exact top. The crystal
was collected under the same conditions as Fig. 48.
-------
B. UNUSUAL CRYSTALS IN ICE FOG (POLYHEDRAL ICE CRYSTALS)
We have noticed many ice-fog crystals have abnormal shapes such as
spherical and near-spherical irregular. Except for ice-fog crystals, we
have never yet observed spherical ice crystals smaller than several hundred
microns in the atmosphere. Since Thuman and Robinson (1954) first dis-
covered the spherical ice crystals (which they called "droxtals") spherical
ice crystals are well known in ice fogs. In the present research, we
found some other shapes of ice-fog crystals different from the normal
sphericals, columns, and plates etc. Figure 48 shows such a crystal with
various focusing or observation angles. Since the crystal was suspended
in silicone oil, the circle and two small dots apparently in the crystal
may not be nuclei, but air bubbles in oil. Another picture (Fig. 49) or
a crystal appears at first to be similar to a normal thin hexagonal plate,
but if we look carefully at the edges of the crystal it is different from
a regular hexagonal plate. This picture is probably a view from the top
(along the c-axis) of the same type of crystal. Figure 50 shows the shapes
of the crystals of Figs. 48 and 49 as drawings. Figure 51 shows a similar
crystal but with a smaller hexagonal top. These crystals seem to be 14-
or 20-faceted polyhedral crystals which have two hexagonal faces at the
top and bottom, 2 sets of 6 trapezoidal faces (total 12) and 6 rectangular
faces between the 2 sets of trapezoids (sometimes the rectangular faces
are missing). Figure 52 is another similar crystal.
These 14- or 20-faceted polyhedral crystals were found in ice fog
at the MUS site in Fairbanks at a temperature of -47C. The finding of
87
-------
Fig. 48. A polyhedral ice crystal found in ice fog at the temperature
of -47C. The crystal was collected at the MUS site on Jan. 3, 1969.
Three pictures show the crystal with different focussings. This
crystal may be 14- or 20-faced polyhedral crystal. The crystal was
suspended in silicone oil.
-------
Fig. 50. Drawings of the crystals in Figs. 48 and 49.
-------
tig. 51. Another polyhedral ice crystal in ice fog under
same conditions as Fig. 48.
Fig. 52. An ice crystal similar to above polyhedral
crystals under same conditions.
-------
these crystals supports strongly Kobayashi's (1965) laboratory experiment,
in which he found pyramid faces at an edge of a prism crystal at temper-
atures between -45 and -55C. The crystal we found in ice fog appears to
consist of 2 (0001) hexagonal faces, 2 sets of 6 (1011) trapezoidal faces
and 6 (1010) prism faces. Sometimes the last 6 prism faces are missing.
Therefore the crystals we found are 14-faced bi-pyramid crystals and/or
20-faceted bi-pyramids with short prism.
Although we do not have good evidence, it is possible that the
formation of these crystals can be attributed to the fact that the super-
cooled water droplets freeze so rapidly that the crystal does not have
enough time to reach.an equilibrium state for the development of normal
hexagonal and rectangular faces. Shimizu (1963) found long prism crystals
lOOu to 1 mm long with the ends probably (0001) faces in the precipitation
in Antarctica. In our case we did not have long prism faces, but we dis-
covered some other shapes of ice-fog crystals which do not have well-defined
faces, such as the crystals indicated as B (for block ice) in Figs. 53
and 54. The crystals found by Shimizu in Antarctica may not have been
formed by freezing of water droplets, because they were found in pre-
cipitation and were too long to be formed from freezing of water droplets.
And in Kobayashi's experiment the crystal must be grown on a certain support
in his cold chamber, under which conditions the crystal develops into a
columnar prism with pyramid faces on an end of the prism.
91
-------
Fig. 53. Ice-fog crystals collected under same conditions as Fig. 48.
Mark A shows a thin plate with two half hexagonal thin plates. Marks
B illustrate "block ice crystal". C2 is a plate crystal which has
another plate angled 60° or 90°.
Fig. 54. Ice-fog crystals taken under same conditions as above. Cj
shows a half plate and a vertical plate attached with it. The weather
conditions were the same as above.
-------
0
L_L
J I
Fig. 55. Drawings of the crystals A and C^ of Figs. 53 and 54.
-------
Figs. 56 and 57. Block ice crystals. The crystal in Fig. 56 was
collected under the same conditions as Fig. 48. The crystals of
Fig. 57 were sampled on December 23, 1965 at the MUS site with
temperature of -40C. The crystals in Fig. 57 were suspended in
formvar liquid.
-------
The possible explanations for our polyhedral crystal are as follows:
1) Under low temperatures, a quasi-stable state can exist for growth of
ice crystals toward the most stable state of ice because of a scarcity
of thermal energy, A trapezoidal face is a quasi-stable surface which
has higher free energy than the hexagonal and rectangular faces in the
case of ice. Under rapid cooling, the growth of a quasi-stable surface
of ice may become comparable to the completion of stable crystal faces
which have lower energy. When the freezing of water droplets in a
spherical shape is complete, the crystal is a little warmer than ambient
air so that water molecules can move over the surface of the ice to form
the quasi-stable surface from the spherical surface. However, the movement
of molecules would be too slow to form the most stable crystal faces after
the trapezoidal surfaces were completed because the temperature of the
crystals has been lowered enough. 2) Due to absorption of air pollution
(dusts inside or outside), the growth did not occur onto the normally most
stable faces [(0001) or (1010)].
We could see such shapes of ice-fog crystals only in the largest
crystals because of the poor resolving power of the optical microscope.
However, these detailed shapes of ice crystals could exist in smaller ice
fog crystals which have been considered as spherical.
In Fig. 53 many block shaped ice crystals can be seen. Among these
a crystal (probably a hexagonal plate) marked as C_ has an additional half
a plate which seems to grow with an angle of 60° or 90°. Figure 54 has
95
-------
also a thin half hexagonal plate marked as C. with an unknown edge, which
might be at an angle of 90° with the basal plane. Figure 53 has another
example of a crystal having a plate angled at 90° or 60° to the basal
plane and the hexagonal plate ice crystal marked as (A) has somewhat
smaller half plates aggregated or grown on the basal plane with angles
of probably 60° each. See a sketch of this as a model in Fig. 54. In
Figs. 55 and 56, many ire block crystals can be seen. The mechanism of
formation of these block crystals can also be attributed to the rapid
freezing of water droplets.
96
-------
8. MECHANISM OF ICE FOG FORMATION
Normal fog or cloud droplets are formed when the relative humidity
rises to more than 100% (but usually not more than 101%) in the atmosphere.
Moisture in excess of that needed for saturation at the temperature involved
condenses onto appropriate condensation nuclei. Thuman and Robinson (1954b)
measured humidities but they did not observe or mention how the humidity
changes during the development of ice fog. Benson (1965) thought that
after the relative humidity reached 100% with respect to water, water drop-
lets would form and then freeze. After the freezing of droplets the
humidity should remain at ice saturation. However, there was no obser-
vational and theoretical evidence for such an explanation at that time.
Thuman and Robinson (1954a) stated in their report that
"several boiler houses and various steam-line vents were probably
the primary sources of moisture for ice-fog aerosols. It is
postulated that, at temperatures below -30C, droxtals and thus
ice fog arise from the direct freezing of supercooled droplets
that have been formed by condensation in moist gases discharged
in the atmosphere. At these temperatures the moist gases can
quickly saturate the air, so that the ice aerosol can persist."
Kumai and O'Brien (1965) reported observations by collection of particles
in silicone oil of ice crystal nucleation process in exhaust gases coming
out of a power plant stack, although the observation was made at -22C.
They also confirmed that water droplets were formed in exhausts from cars,
heating and power plants, and it was suggested that the ice-fog crystals
result from freezing of the supercooled water droplets. They did not give
further conclusions. Benson (1965) stated, concerning the mechanism, that
the rapid cooling condenses the water vapor into small droplets which be-
come supercooled, and at temperatures below -30 to -35C they freeze.
97
-------
These three most comprehensive studies of Fairbanks ice fog have
shown neither evidence nor detailed observation and theoretical consider-
ation for the freezing of water droplets. (Kumai and O'Brien's experiment
was carried out at higher temperatures than those at which ice-fog appears,
and they used silicone oil which might freeze the supercooled droplets
when they are collected in oil) . From the observations we made in the
present research some evidence on the freezing of water droplets has been
given in the previous chapters in describing the positions of ice-fog
nuclei and the appearance of spicules in ice-fog crystals. In this
chapter more detailed observations and theoretical studies from various
viewpoints will be discussed.
a. Measurements of Humidity under Low Temperature Conditions:
It is well known that it is difficult to measure the humidity of air
in the temperature range below -30C, although several methods have been
developed. Even for temperatures between OC and -30C data are scarce
(Munn, 1966).
The traditional method of dry and wet bulb thermometers gives
questionable accuracy at near and more than ice saturation environment.
An attempt at humdity measurement of warmed samples of initially cold
air by use of the dry and wet bulb thermometers has been proposed by
Branton (1965). However, for air at temperatures of -40C, the difference
in readings of dry and wet bulb thermometers is several cens of degrees
when the air has been warmed to above freezing temperatures. A dew,point
hygrometer is good for measurement of humidities lower than ice saturation
for sub-freezing temperatures, but for ice fog studies we also need
98
-------
accurate humidity measurement between ice and water saturation for
temperatures as low as -55C. A "dew cell" cannot be used for temperatures
lower than about -30C. A carbon coated sensor for the determination of
humidity is also being used at the present time, especially in radio-
sondes. However the electric resistance of the carbon humidity sensor
is also changed by temperature variation, and at -40C the resistance
becomes unmeasurable according to Stine (1965) and the author's trial.
Using infrared or electronic radiation we could probably measure both
total water content and water vapor in the air with fair accuracy. However,
at the present time the equipment for these methods is too expensive and
requires too much time. Also there is no guaratee that the machine would
operate well at the low temperatures involved. Traditional hair or mem-
brane hygrometers also do not work below about -25C, because the structure
of the hair or membrane will be broken by freezing and a serious time lag
develops. However, if the hair is treated with BaS to avoid the hysteresis
effect, is rolled with a special technique, and is treated with H So the
accuracy of the hair hygrometer becomes better then +^ 2% in relative
humidity (Kobayashi 1960), probably even for -40C. It is possible that
a recording hygrometer with the specially treated hair might work at
lower temperatures.
i. Data from Hair Hygrometers:
As was previously mentioned a hair hygrometer would not work properly
in low temperatures due to the low time lags. However, average values
for a fairly long time interval by use of the specially treated hair may
99
-------
be valid, even though it would not give very accurate values of instanta-
neous humidity. Fortunately, the weather of interior Alaska in the winter
is so steady that sudden changes of humidity probably do not occur except
during a very few cases of storms. Therefore, the data obtained from hair
hygrometers may be useful for a comparison of humidity trends between
different locations (with and without fog). Using these records,
the humidities recorded at 0000 AST and 1200 AST from the middle of
November 1967, through the middle of February 1968 were plotted against
temperatures (see Fig. 58). The figure shows that humidities measured in
the ice-fog area (the MUS site) are a little greater than those measured
in the area free from ice fog, especially below -20C. This implies that
the humidities in the ice-fog area are always affected by the local sources
of water vapor such as power and heating plants, automobiles and open
water. It is important, however, to note that there is not an extreme
difference between humidities in ice-fog areas and areas free from ice
fog. Furthermore, it is also important to note that water vapor pressures
near the ground lie on the average between ice - and water - saturation in
the temperature range below -20C in both the ice-fog areas and the areas
free from ice fog.
The record of humidities at the Ester Dome site sometimes showed
lower humidites than ice saturation and distinct variation even below -30C
(the minimum temperature recorded at Ester Dome in the winter of 1967 and
1968 was -34.4C). On the contrary the records at the MUS site and at the
Geophysical Institute site did not show such humidity variation below
-30C. These observations imply that above an ice-fog layer, humidities
100
-------
ESTER DOME
90
70
*>5
2j50
|= -40 -30
-20
rIO -
cc
UJ90
o
570
13
I
UJ
UJ
50
GEOPHYS. INST.
-30
-20
-10
0
70
50
MUS
•at*.
J L
J L
J L
:40 -30 -20 -10
TEMPERATURE (°C)
0
Fig. 58. Relationship between relative humidities provided by hair
hygrometers and temperatures during the winter of 1967
to 1968 at various sites.
-------
are affected by the air coming from distant areas, but in an ice-fog layer
the air is stagnant. Also since the temperature and humidity records show
almost simultaneous changes even below -30C (at least down to -34.4C) the
hair of our hygrometers may have already been treated with BaS and H_SO,
and rolled and should probably work at temperatures below -40C. The
humidities recorded by our hygrothermograph (model H-311, Aerojet-General
Corp.) at the MUS site were approximately equal, on the average, to those
measured by the absolute method which is used to calibrate the hygrometer
and which will be described in the next section.
ii) Absolute Method of Humidity Measurement:
Thuman and Robinson (1954b) measured humidities in the temperature
range between -20 and -43C in ice fog, using the Karl Fischer reagent
method, which operates as follows: First a measured volume of air
is drawn through absolute methanol which is used as a water extractant,
and second, the amount of water dissolved in the methanol is determined
by titration of the Karl Fischer reagent. However, this method requires
considerable skill and also time for the determination. In addition to
the above, there are some doubtful points at the measurement of amount of
air drawn through the methanol and the ice-saturation reference. In
order to measure the air volume, they used a test meter. However, in the
low temperature below about -20C, a test meter such as that used by
Thuman and Robinson seems to be improper for measurements of flow rates,
because the inside of such a meter does not respond accurately enough at
the low temperature to changing flow rates. Nakaya (1954) used a method
102
-------
similar to theirs, but he used a drying tube containing P 0 (phospheric
.pentoxide) which was the same material used by Tyndall (1922) and
Hanajima (1944) as an absorbing agent of water vapor. Nakaya determined
the amount of air through the tube by means of an evacuated 20 liter
glass bottle, allowing the air to come slowly into the bottle until the
pressure inside of the bottle reached barometric pressure. Even though
the method works well at around -20c, for -40C or below 20 liters of air
does not contain enough water vapor to give a measurable weight difference
in the ^2^5 tu^e> *n addition, operation of a vacuum pump at such low
temperatures creates problems. Finally, we had mechanical problems with
the barometer used to measure barometric pressures at the various sites
in ice fog.
In the present research we used an air sampling method similar to
the previous methods» We used magnesium perchlorate, Mg(C10,)9 as an
absorbing agent, as was suggested by Kumai and O'Brien (1964). Other
than magnesium perchlorate or phosphorus pentoxide P2^s» t*xere are many
kind of water absorbing agents such as sulfuric acid (H?SO,), calcium
chloride (CaCl_) etc. The magnesium perchlorate was chosen as the most
effective and convenient absorbing agent. Also calculations of the expected
temperature rise in its use for humidity measurement under ice-fog
conditions give a negligibly small heat of reaction. To measure the volume
of air, methods such as the use of a test meter or flowmeter cannot be
applied because of inaccuracies caused by low temperatures. We had
considerable difficulty in trying to use the flowmeter for the measurement
103
-------
of air flow under ice-fog conditions with pressure and temperature
corrections. Since the flowmeter was connected behind the drying tube,
the pressure at the head of flowmeter was much different from 1 atmospheric
pressure. Using a pressure gauge also gave us very inaccurate values.
Mercury to measure pressure has also been frozen. Temperature correction
for the original calibrated scale in use under ice-fog condition is not
insignificant. Furthermore, we had to measure accurate barometric pressure
and temperatures at the same locations simultaneously.
After wasting a lot of time, we finally used two 50 liter glass
bottles filled with kerosene and connected by tubes through an electric
pump (normally used for water transfer) to measure the amount of air.
This seems to be the best technique developed so far for measuring the
air volume at temperatures lower than -20C. Kerosene neither freezes nor
bubbles even below -50C and the pump (with a brush-type motor) operates
at ambient temperature below -40C without any difficulties. Antifreeze
liquids for automobiles also do not freeze but when they are transferred
from a bottle to another bottle, bubbles develop in the bottle, so that
the measurement of volume becomes inaccurate. As indicated in Fig. 59,
two glass bottles are used to measure the volume of air. It is better
to use a larger bottle, because the accuracy of the measurement increases
with the weight of the drying tubes. But for the convenience of trans-
portation for the device, 50 liter bottles were used.
In order to filter out solid water particles, a 0.45 micron pore size
Millipore filter is used. This type of filter is a plastic, porous
104
-------
MILLIPORE
FILTER
DRYING TUBE
7 • • • SCALE
KEROSENE PUMP
Fig. 59. Arrangement of apparatus for humidity measurement.
0.9 r
-35 -30
TEMPERTURE (°C)
-25
Fig. 60.
Water vapor density versus temperatures under ice-fog
conditions at the MUS site. The curves for water- and
ice-saturation are according to Smithsonian Meteorological
Table.
-------
structure, and has a low resistance to the flow of air. It was confirmed
that the filters do not absorb or release any appreciable amount of water
vapor under the conditions of low water vapor pressure and that the visible
deposits of ice-fog crystals on the surface of filter during suction of air
through the filter give negligible error.
The procedure of measurement is as follows:
1. The dry weight of the drying tube containing magnesium perchlorate
is measured with a chemical balance which has a precision of
+ 0.01 mg.
2. The drying tubes, which are stored in a desiccator, are kept at
ambient (outdoor) temperature until measurements are made.
3. The drying tube is connected to the rubber hoses, both glass
stopcocks of the drying tube are opened, and the pump is switched
on. To handle any part of the drying tube clean dry tissue
paper should be used.
4. Air is taken into the bottle through a 0.45y pore size filter
and a drying tube. After 30 liters of kerosene are passed to
the second bottle, the pumps are stopped. The volume of air
which passes through the drying tube is measured by a scale on
the kerosene-filled bottle.
5. The drying tubes are weighed on the chemical balance after the
tubes are warmed up to room temperature. If the temperature of
the tubes is lower than room temperature, water vapor condenses
on the tubes, thus causing error.
106
-------
6. The water vapor density A is given by
A - X. (gm m"3),
V
where w is the difference of weight in mg before and after
absorbing the water vapor and V is the volume in liters of air
taken. Using the value A, water vapor pressure e in mb can be
calculated by the equation
e =( AT - (mb),
where T is absolute temperature. R is the gas constant, e is
m /m. (= 0.622), m and m, are molecular weights of water and
w a w d
of dry air, respectively.
Relative humidities over water, U. and over ice, 43. are easily
calculated from e divided by che saturation vapor pressure over water
E and ice E , respectively, from the Smithsonian Meteorological Tables.
However, the values of E and E. in these tables for temperatures below
w i
OC were extrapolated from those at temperatures from OC to 100C and are
pending further research, according to the table.
A rough estimate of the error in vapor pressure is about 1 percent
of value observed. However, care must be taken that the air is not passed
through the drying tube too quickly because the water vapor cannot be
fully absorbed by magnesium perchlorate at high flow rates. In order to
check this two tubes were used in series. However, the second tube showed
107
-------
no sensible increase in weight. Hence, throughout the observations, only
one tube was used for each measurement. Measurements were made at a
flow rate of 10 liter per minute.
In order to check the reliability of the data obtained, we measured
simultaneously the humidities with an Assman type hygrometer in the
temperature range OC to -IOC and then compared the values. The two values
agreed to within 1 to 2 percent in units of relative humidity. (See
Table 8). An Assmann type hygrometer, in general, shows accurate values
to several degrees below the freezing temperature.
TABLE 8
Comparison of the Data on Humidities between
The Method Used h.ere and Assman Type Hygrometer
Temperature
-1.8
-5.4
-6.9
-7.2
-8.0
Absorbing
Method
65.0 percent
75.0
49.4
77.3
49.4
Assmann
Hygrometer
65.2 percent
74.0
48.0
76.0
50.0
The water vapor content was observed at the MUS site during the
ice-fog events of the winter of 1967-1968. The results appear in Fig. 60,
which shows the relationship between temperatures and water vapor densities.
In the figure two curves illustrate water saturation and ice saturation
according to the Smithsonian Meteorological Tables. From the figure, it
108
-------
is obvious that generally the water vapor density under ice-fog conditions
lies between ice- and water-saturation, assuming the values for E and
w
EA are accurate. Assuming accurate values of E and E on the figure,
we can express the water vapor content in terms of relative humidity.
Thuman and Robinson (1954b) reported that the average value of relative
humidity with respect to ice was 91 percent under ice-fog conditions. The
average humidity obtained in the present study is 23 percent higher than
this. The reason for this difference cannot be found at the present stage.
However, since ice-fog cystals stay stable or grow very slowly under
conditions of persistent ice fog in the central area of the city, the water
vapor density must lie between ice - and water - saturation, as we obtained
experimentally.
b. Visual Observations of Ice-Fog Sources:
On the basis of careful observations of ice fog from the ground and
the air, we can say that the densest ice fog originates from open water
such as cooling water dump areas from power and heating plants, while
relatively thin but widely spread ice fog comes from automobile exhausts.
Exhausts from power plants and commercial and residential heating plants
also make ?'-ie fog. Figure 61 is a photograph* of the Fairbanks area taken
i
by Mr. Mob ley of the National Geographic Society from the air on a morning
in November 1968. Although the exact time and ground temperature are not
The original color picture is available in "Alaska", published by National
Geographic Society, Washington, D. C., 1969.
109
-------
Fig. 61. An aerial photograph of Fairbanks taken just before ice
fog forms. (Photo by Mobley, National Geographic Society).
-------
known, the picture appears to have been taken around 10 a.m. and at a
temperature of about -25C (which is a little warmer than the temperature
at which ice fog appears). The picture splendidly displays some of the
major sources of moisture which produce dense ice fog. The thick steam
shows up clearly along the slough which is kept open by cooling water dumped
by the Golden Valley Electric Assn., but the steam from the right fore-
ground of the picture to the foreground bend of the slough seems to be
evaporating before freezing, probably due to a shortage of moisture and
freezing nuclei in this area. However, just beyond the second bend of
the slough there is a thin layer of ice fog spreading out from near the
actual dump point below the large plume of the Golden Valley generating
plant, and there is a long layer of ice fog above the ground. It is
obvious that the exhaust from the stack of the MUS power plant (on the far
right side, of the picture) goes up and dawn in the city, dissolving and
changing into thin ice fog. Many small stacks of home or store heating
are also making .ice :og. Beyond the Golden Valley Electric Association
plant, the Fort Wainwright power plant and its cooling pond are also making
he,'vy ice fog. Along the road which can be seen from the lower right hand
corner to the middle left of the picture, several cars are running, but
their contribution to formation of ice fog does not seem to be significant
in this picture. Even though the picture does not show true ice fog, the
moisture sources for ice fog are identical.
Along highways we normally can see dense ice fog and it seems to form
from automobile exhaust. However the contribution of car exhaust may be
111
-------
over-emphasized by many people. In fact, during the severe cold spell
from the end of 1968 t" rough the middle of January, 1969, dense ice fog
around the International Airport of Fairbanks originated from the main
airport building and several attached cargo, post office, oil service,
and local air-line offices rather than from automobile exhaust in the
parking lot of the airport. Under the -50C cond'tion on 2 January 1969,
the visibility leeward of the building was about 100 m but between the
building and .leeward of the parking lot was 120 m, and windward of the
parking lot the visibil'iy was also 120 m. The poor visibility windward
of the parking lot was due to the small amount of open water west of the
Bureau of Land Management office (BLM) and the University power plant
(see Fig. 2) and/or coming from the residental area located just northwest
of the airport. At any rate, from our visual observations, car exhausts
were not the most important source of water vapor for ice fog. However,
Concerning air pollution in general, we agree with Benson's (1965) con-
clusion that car exhausts are very serious sources of such components of
air pollution as hydrocarbon materials and lead compounds, although we
did not have definite evidence for these materials.
Exhausts from power plants also make ice fog under low temperature
conditions*. The smoke from a stack initially rises higher than the top
of the ice-fog layer because of its warmer temperature, but the plume then
cools and sinks back down into the fog layer. Leeward of the stack we
could see wide spread diamond dust (= ice crystal display) when ambient
— - _ . _ _ .
This description is revised from the former progress report entitled
"Alaskan Ice Fog" (Ohtake, 1967).
112
-------
temperatures were lower than -21C. The particles of the smoke were con-
firmed to be a major source of ice-forming nuclei effective at temperatures
below -13C with a high degree of supersaturation (Ohtake and Henmi, to be
published).
Observations of the development of ice fog in the Fairbanks area
from its beginnings also showed that at first the fog develops from open
water and spreads out into the area downwind from the open water, as can
be seen in Fig. 61; then at temperatures below -21C ice-crystal displays
appear downwind from power plants as well as open water. Under these
conditions visibility is reduced to about 1 km. As temperatures continue
to drop to about -30C, slight ice fog appears in the vicinity of houses
and along the roads. At temperatures lower than -35C the ice fog completely
covers the city. Although ice fog from open water is dense, it does not
seem to spread out over very wide areas (the exact area varies with the
direction of the wind), while ice fog originating from automobile exhausts
was thin but widespread.
The onset temperature of ice fog has been gradually increasing. In
1911 there was virtually no ice fog even at -50C, even in Fairbanks city
as shown in the photograph reproduced by Benson (1965) as Fig. 20. Benson
(1965) gave -35C as the upper temperature limit for ice-fog formation, while
our own more recent observations suggest that the current upper temperature
limit for ice fog is nearer -30C. With the current 'oil boom', the pop-
ulation, number of houses and cars, and airport traffic are all increasing
113
-------
rapidly. Figures 62 and 63 show the relationships between temperatures
and visibilities at the Fairbanks International Airport for the winters
(December through February of the next year) of 1964 to 1965 and of 1968
to 1969, respectively. The plots were taken from 3-hourly data which
appear in the Local Climatological Data of Fairbanks published by the
US Weather Bureau. F~om the figures it can be seen that the visibility
in recent years has become worse than that of several years ago at the
same temperatures. Also the onset temperature for ice-fog formation has
risen. So the onset temperature of ice fog is becoming steadily higher,
and this tendency will continue in the future unless a deliberate effort
to reduce ice fog is made. Air pollution other than ice fog (or rather
the pollutants other than water vapor) will also be more serious and the
visibility will continue to deteriorate due to higher concentrations of
ice-fog crystals.
The airplanes at the airport are also producing ice fog, especially
during take-off. But as far as we are able to observe at the Fairbanks
International Airport, such ice fog normally drifts with the wind at
about 2 m sec toward the south to southwest, away from the airport.
However the buildings incidental to the airport are making more serious
ice fog on the runway than are the airplanes themselves. Again, the oil
activity on the North Slope has resulted in a tremendous increase of
buildings adjacent to the runway.
114
-------
-60
100
TEMPERATURE (°F)
•50 -40
-30
10
CD
CO
> I
0.
DEC 1964-FEB 1965
-50
-45 -40
TEMPERATURE (°C)
-35
Fig. 62. Relationship between temperatures and visibilities at the
Fairbanks International Airport during December 1964 through
February 1965.
CO
UJ
-I 5
.3/4CO
'16
-------
-60
TEMPERATURE (°F)
-50 -40
-30
DEC 1968-FEB 1969
10
>•
m
CO
>
0.1
WINTER OF '64 '
-6
-50 / -45 -40
TEMPERATURE (°C)
-35
Fig. 63. Relationship between temperatures and visibilities at the
Fairbanks International Airport during December 1968 through
February 1969. Compare with the situation as of four years ago.
10
CO
LJ
3/4 CO
'16
-------
c. Measurement of Evaporation of Water from the River or a Pan
and the Formation of Water Droplets from the Vapor:
Benson (1965) estimated the amount of water vapor evaporated from the
o 2
open water area of about 1.5 x 10 cm maintained by dumping hot water from
the MUS powe. plant into the Chena River at rate of 482 x 10 g day"1
during clear weather and 522 x 10 g day" for ice fog weather. (The in-
creafc, may be due to increased power demand in colder weather.) These estimates
give evaporation rates per unit area of the surface of the river at OC of
_2 -i _2 -I
32,100 g m day and 34,800 g m day , respectively. Since he estimated
these values by means of the heat budget of dumped water, and his values of
evaporation under low temperature conditions were rather high (of the
order of 30 mm day ,) we tried to observe these values directly. Benson
reported that the largest source of water vapor for ice fog comes from open
water along the river, sloughs or ponds. In the present research we concen-
trated our attention on the open water as a main source of ice fog. On the
other hand we could not find any observations of evaporation rates under such
low temperatures.
An attempt has been made to observe the evaporation rate by using shallow
pans and a heat lamp under ice-fog conditions. Two shallow pans (13.4 cm and
30.5 cm diameters with depth of 1.7 cm: lids of 35 mm film cans) were filled
!
with about 1 cm of water, put on a wooden box and exposed to the atmosphere.
The water was kept free from freezing by use of an industrial infrared lamp
(infrared and visible light). The resultant temperate < of the water were
117
-------
kept around 4 to IOC, with an occasional maximum of 15C. The temperatures
were measured by dipping small thermometers into the water and we tried
to keep them around 4 to 6C by controlling the distance between the lamp
and pan. As can be seen in Table 9, the amount of water evaporated in
-2 -1 -1
ice fog was roughly 5,000 g m day (or 5 mm day ) which is much smaller
than the value Benson estimated. This value indicates that the production
_ -I f\
rate of water droplets is about 101* droplets sec from 1 cm of water
surface assuming droplet diameters of 10y. This value implies that 100
droplets are continuously supplied into the atmosphere from OC water
surface when the strength of wind is assumed to be 1 m sec . In com-
parison with evaporation rates in a room at 23C (water and air temperature)
much more water evaporated from a pan in the colder outside than in the
room. This result is of course, due to larger air movement on the water
surface and larger temperature differences between air and water in the
outdoor conditions. Thompson (1967, unpublished) also estimated an
_2 —i
evaporation rate of about 2780 g m day using Thornthwaite and Holzman's
equation (see Munn, 1966) with a steady-state assumption (no large-scale
transport of fresh air coining into the river area and an adiabatic lapse
rate) .
For constant water temperature, lower air temperatures should give
higher rates of evaporation, but we could not verify this because of
difficulties in keeping the x^ater temperature constant. However an
-2 -1
evaporation rate of about 5000 g m day for air temperatures of around
-40C should be valid under normal ice-fog conditions.
118
-------
TABLE 9.
Observation of Evaporation Rate of Water from Pans in Low Temperature
November 17, 1967 Room Temperature 23 C
Time AST
0930
Total Weight
300. OS™ start
1036 299.0
1212 297.7
1342 296.6
1702 294.3 end
During 447 min. (0930 1702), 5.7 gm of water evaporated.
This value gives 1300 gm" day'1.
Time Date
0930-17 Nov. 1702-17 Nov. 1967
1445-26 Nov. 0906-27 Nov.
0910-27 Nov. 2007-27 Nov.
1700- 5 Dec. 0900- 6 Dec.
0900- 6 Dec. 0900- 7 Dec.
1130- 7 Dec. 0830- 8 Dec.
0900-13 Jan. 0900-14 Jan. 1968
0900-14 Jan. - 0900-15 Jan.
0900-15 Jan. - 0900-16 Jan.
0900-16 Jan. - 0900-17 Jan.
0900-17 Jan. - 0900-18 Jan.
0900-18 Jan. 0900-19 Jan.
Mean
Air
Temp.
23
-23
-18
-30
-25
-39
-41,
-42
-44
-43
-43
Water
Temp.
23
(2) w/ice
(-) wo/ice
4
7
6
5 ^ 10
10
15
15
4
5
Small Pan
-? -i
gm day L
1300
2890
3190
3200
5700
5460
5580
5170
8530
7160
3110
4530
Water
Temp.
-
4
6
7
11
10
6
5
Large Pan
gm~ day"!
4440
5320
6860
6110
5460
7370
7560
4280
4350
119
-------
d. Temperature Measurement above Water Surface:
The main sources of moisture in the formation of ice fog are open
water and exhausts from automobile, heating and power plants. The temper-
atures of car exhausts were measured by Benson (1965) , and he gave an
empirical formula
= _k (T - T )
dx k tT V
where T is the temperature of the exhaust gas (°C), T is the ambient air
ŁL
temperature, x is the distance (cm) from the exhaust tube outlet, measured
in the center of the exhaust plume, and k is a constant. We obtained
similar data (Ohtake, 1966) and the equation seemed to fit well.
Open water is maintained where power plants dump info the river a
lot of cooling water, which keeps some of the river from freezing up. Since
the temperature of this open water on the Chena River is close to OC, even
in ice-fog condition evaporation is high and the resulting steam fog drifts
into the city of Fairbanks. Measured temperature profiles above the water
surface using a thermistor probe at the end of a long pole are shown in
Fig. 64. These temperature profiles seemed also to fit Benson's equation
for car exhaust. With this steep temperature gradient and slight air
drainage winds (about 1 to 2 m sec ) blowing down from both sides of the
banks, the steam rises and is cooled very rapidly. Kumai and O'Brien
(1965) measured the temperature downwind of a power plant stack at the'
height of the stack top. Their results seem to be reasonable.
120
-------
o
o
^*H*
^
l_
1
<
a:
UJ
Q_
-10
-20
-\ \ *•
\ %»\
» %c
» <
»
\
\
\
«
-30 -
CHENA RIVER
-40
I 5 10 20 50 100 200
HEIGHT ABOVE WATER SURFACE (CM)
Fig. 64. Temperature profiles above open water under the conditions
of several ambient air temperatures.
io
-2
O -:
a: o
X UJ
o
Si
O o
I- o
-J Z
00
10
-3
3
O
cr
UJ
1°
Si
"S
O Q
-2
UJ
-50 -40
AMBIENT TEMPERTURE
-30
Fig. 65.
Result of calculation of time for water droplet to cool from
OC to -30C through conduction cooling or radiation cooling.
-------
e. Derived Supersaturation Degrees, Compared with Supersaturation
Required to Activate Condensation Nuclei:
From the previous section, we know that the temperatures above the
open water are almost the same as the ambient air temperature more than
a few centimeters above the water surface. Exhausts from automobiles and
power plants also cool very rapidly. Air over the open water is easily
moved by slight winds and small scale convection to a few centimeter above
the water surface, where it encounters extremely cold air. If the air is
saturated at the water surface, the vapor pressure is 6.11 mb (or saturation
_3
vapor density is 4.85 g m ), while saturation vapor pressure over water
_3
may be 0.19 mb (or 0.18 g m vapor density) for -40C. Even at -30C
_3
the corresponding values are 0.51 mb or 0.45 g m . If the saturated air
at OC cools to -40C in 5 cm, we will have a factor of 30 times saturation
(relative humidity 3000%). Even if the ambient air is considerably warmer
the supersaturation will be very large. At these high supersaturations,
water vapor from open water or combustion exhausts can very easily condense
into water droplets. If water temperatures are higher as at hot springs,
the supersaturation will be even higher. As was stated in the previous
chapter, at Chena Hot Springs and Manley Hot Springs concentrations of
condensation nuclei effective at about 400% of saturation were normally
—3
200 to 400 cm . Even at such low concentrations of nuclei, the super-
saturation was high enough that condensation of water vapor into water
droplets can easily occur. In the Fairbanks area where we can find more
5—3
than 10 cm condensation nuclei, condensation will be even faster. This
is quite a different situation from that in natural clouds, where such
122
-------
supersaturations would never occur. Only in a contrail are conditions
similar. At low temperatures, automobiles and heating and power plants
produce water droplets in the air in the same manner as does open water.
Even though available dusts must be effective as condensation nuclei in
areas where nuclei are scare, it is very possible that fog droplets form
without any nucleus,
f. Time Required for Droplets to Freeze:
Some water droplets may evaporate and the water vapor added to the
atmosphere may provide moisture for growth of ice-fog crystals, but this
evaporation of droplets does not seem to occur to any great extent in the
vicinity of the water surface. If much evaporation occurs, the latent heat
of evaporation has to be taken from the droplets themselves, so that
freezing proceeds in the droplets. Thus the water droplets are cooled
down and frozen into ice crystals. Once the droplets freeze, the crystals
cannot evaporate so quickly. They then drift into the ice fog. Exhaust
from cars and heating plants react similarly except that these exhausts
may produce nuclei as well as water vapor.
In order to explain the transition from a water droplet to a spherical
ice-fog crystal, it is necessary to know the time required for water droplets
to freeze. Two mechanisms for cooling of droplets, conduction and radiation,
are considered in this report.
1. Conductive Cooling:
Since a water droplet is small, the temperature in it is assumed
homogeneous, assuming that temperature T(r) in the air satisfies Poisson's
2
equation, V T(r) = 0,
123
-------
dT
-4ir r k (T-T ) = m C -j-
a o dt
where
r : distance from the center of droplet
r : radius of droplet
Ł1
k : coefficient of thermal conduction in air
T : temperature of droplet
T : ambient temperature
m : mass of droplet
C : specific heat of water
The time required for water droplets of 12v diameter initially at
OC to cool down to -30C is shown in Fig. 65. The time is proportional
to the reciprocal of the square of the diameter of the droplet.
2. Radiation Cooling:
The fundamental equation is
-. f (aX - T*> -.of
where
S : surface area of water droplet
T : radiative temperature of upper hemisphere
T : radiative temperature of lower hemisphere
Af
a : Stefan-Boltzmann's constant
The measurements of radiative temperature for several objects were
made by use of a Linke-Feussner radiometer in the ice-fog condition, as
124
-------
shown in Table 10 including data of Stoll and Hardy (1955). On the basis
of the observations, the following cases may be considered:
i) Droplets above the water surface and without fog; T = 213°K
and T. = 273°K.
%
ii) Above the water surface and with fog; T = ambient temperature
and TŁ = 273°K.
iii) Above the snow surface and without fog; T = 213°K and
T = ambient temperature.
iv) With fog and above snow surface; T = T = ambient temperature.
11 JO
We now calculate the time for the droplet to cool down to -30C. In
cases i) and ii), the droplets never get colder than -25C regardless of
their size. Case iii) gives results similar to those for case iv). There-
fore we will show only the result for case iv). The form of the curve of
time versus ambient temperature is approximately the same as that for
conduction cooling, though the time is now proportional to the reciprocal
of the diameter. Therefore, we need only change the scale in Fig. 65 to
show the result.
TABLE 10
Radiative Temperature of Various Objects
Sky
Sky
Sky
without fog (after Stoll and Hardy, 1955)
with thin cloud
with ice fog
Snow Surface
Water (OC) through 20m thick of ice fog
Air
Temperature
°C
-40.0
-35.2
-40.1
-40.0
-40.4
Radiative
Temperature
°C
-65
-41
-40
-36
-20
125
-------
In the above calculation, the OC means the steam has formed just
above a water surface of temperature OC and the -30C means the ice fog
begins to appear about -30C according to our observation. The initial
temperature for droplets originating from exhaust of cars and heating units
may be approximated as 50C to 100C. Even for an initial temperature of
100C, the time to reach -30C is of same order as that for the initial
temperature OC.
As an experimental verification of the above calculation the frequency
of appearance of water droplets near water sources has been examined by
two methods: the collodion or formvar film method and water-blue film method.
Droplets and crystals in the air are captured by very weak suction on a
collodion or formvar film supported by an electron microscope grid, the
temperature of which can be controlled by varying the voltage to the
illumination lamp or by warming the microscope stage. When the air containing
water droplets and ice crystals is sucked through a small impactor (Fig. 66)
by mouth, both kinds of particles hit the film. Ice crystals do not
evaporate during a very long time on the grid, while water droplets hitting
the film do evaporate in about a second. This is due to the difference of
saturation vapor pressure for ice and water, and also the thinnest formvar
or collodion film provide a very small heat capacity which prevent the
deeply supercooled water droplets from being frozen by hitting a cold support.
Unfortunately, we did not succeed in obtaining good demonstration pictures
by use of a microscope and a 16 mm camera. By this method we confirmed;
that some x^ater droplets were still in the liquid phase 5 m downwind from
OC open water at an ambient air temperature of -42C with a wind speed of
126
-------
MICROSCOPE
RUBBER TUBE
wn \
MICROSCOPE
OBSERVER'S MOUTH
STAGE
CONDENSER LENS
(
V
Id'
VOLTAGE
CONTROL
AC 115 V
Fig. 66. A small impactor to check whether the water droplets have been frozen or not.
-------
1.5 m sec . 10 m from the open water few droplets can be found. These
values indicate that the time required for water droplets of about lOw
diameter to freeze is about 3 to 7 seconds at an ambient temperature of
-42C.
Another method is based upon Okita's (1958) water-blue film method
to determine the drop size of water fogs. He used a cellulose nitrate
film coated by a 2.5% aqueous solution of water-blue (same as aniline
blue) dye. If any water droplets contact the film, they produce clear
spots on the blue film base. These clear spots are undoubtedly made by
re-solution of the water-blue coat, but under very low temperature conditions
it is not certain that such water droplets will have time to dissolve the
dye before freezing on the film. However under -42C temperatures and
at nearly the same time as the formvar film method was being done, we
found many droplet spots on the water-blue film which was tested at the
bank (0 m) and 20 cm above the open water. The films were attached to
slide glasses 3 mm wide, which were waved 30 times (about 1 m of traverse)
by hand as fast as possible to improve collection efficiency. At 1 m
from the open water many faint spots and several clear spots were found.
We presume that the faint spots were made when a water droplet hit the
water-blue film, froze and evaporated, while the clear spots were made by
water droplets which evaporated from the film without freezing. On the
other hand, the water droplets (= ice crystals) which are frozen before
impact will not make any spots on the water-blue film. There were small
clear spots and faint spots at 2 m, some faint spots and no clear spots
at all at 5 m and no spots at all in any shape 10 m from the open water.
128
-------
These results are almost the same as those made by the formvar film method.
At the MUS site in downtown Fairbanks and 20 m or more from the hot springs
at Chena Hot Springs, we have never found any spots in several trials.
These observations verify that all particles in ice fogs are completely
frozen (except near water-droplet sources) under conditions of about -35C
or below. This result is not in agreement with Borovikov's (1968) who
observed some supercooled water droplets in a natural cloud at a temperature
of -40.6C, but we could not obtain detailed information about the method
he used.
g. Theoretical Study of the Size Distribution of Ice-Fog Crystals:
Considerable theoretical attention has been given to the .nucleation
and growth of ice crystals in clouds (Mason, 1957; Fletcher, 1962; Byers,
1965). The present study deals with the theoretical treatment of the
nucleation and subsequent growth of ice-fog crystals, based upon the
observations we have made such as the measurements of temperature profile
above water and behind automobile exhausts, high supersaturation near the
sources, absolute humidities, and possible homogeneous condensation and
freezing of water droplets. Since a full paper has been submitted to
"Tellus", by Huffman and Ohtake, and is attached to this report as an
appendix, only the abstract of the paper is given here.
A mechanism is proposed for the formation of ice-fog crystals in the
city and environs of Fairbanks, Alaska. Equations are developed for
calculating the size distribution resulting from growth by sublimation
of water vapor. These equations are solved numerically, with the use
of a computer, for three major types of ice-fog sources: A) auto-
mobile exhaust, B) exhaust from heating plants and C) open water. For
129
-------
source type A, the computed size is much smaller that that observed;
but for source types B and C, the computed size distributions are
found to be in good agreement with experimentally obtained values.
The discrepancy between the computed and observed sizes for source
type A is possibly due to the large degree of supercooling exhibited
by small water droplets. Only very few of the small water droplets,
initially formed by condensation, freeze. These frozen particles may
then grow to the observed size at the expense of the more numerous
evaporating liquid droplets. On the other hand, the larger droplets
produced by source types B and C probably nearly all freeze very
shortly after their formation, leaving no appreciable supply of liquid
droplets to provide for their further growth.
h. Effects of Lower Air Temperature:
It is important to realize that the lower temperatures are not only
responsible for the early freezing of water droplets but also affect many
aspects related to the ice-fog crystals. The rate of evaporation from the
water surface depends upon the differences between water vapor densities
of ambient air and air at the water surface. Since the water vapor density
at the water surface is constant because the water temperature is constant
at OC} the lower temperature gives the higher evaporation rate because
colder air has lower water vapor density. Also, colder air can maintain
smaller amounts of water vapor. Thus the lower air temperatures gives the
higher rates of water-droplet formation even in same wind profile above
water surface, and quicker freezing.
Under the lower air temperature conditions such frozen water droplets
(= ice crystals) can grow only at a minor rate because values of vapor
density differences (p - p ) is smaller in the cold air assuming ambient
/
vapor density is at middle value between water - and ice - saturation
130
-------
(refer to the equation on the page 83 and observation of humidity in ice
fog). This results in many small rudimentary faceted ice crystals which
have very slow fall speeds and are persistently suspended in the air near
the ground for long time periods; thus ice fog is continued. This mechanism
explains that at lower temperature the mean size of crystals becomes smaller
and the concentration of ice crystals becomes higher. Also, the solid
water content should be temperature-dependent, providing that the moisture
source remains constant and water droplets are released from the same area.
With the lowering of temperature, the air loses its capability to
maintain the same amount of moisture as vapor, and steam or ice crystals
which come from open water evaporate only with great difficulty. Such
unevaporated ice crystals make ice fog.
This descriptive explanation was made based upon our observations
for the case of open water, but it analogizes the cases of exhausts from
heating plants and cars in which the temperatures are higher than OC in
these cases.
These low-temperature effects may be more important than the role
of freezing nuclei for ice-fog formation in the inhabited area. From
figures 62 and 63 we could not find any threshold for appearance of ice
fogs at temperatures between -37C to -40C which have been considered to
be the most important temperature for ice-fog formation. If the critical
temperature for formation of ice fogs in relatively uninhabited areas
such as Manley Hot Springs or Chena Hot Springs would be -40C, the ex-
planation of low temperature effects may be more valid than that through
the spontaneous nucleation of purer water droplets.
131
-------
i; Conclusions on the Mechanism of Ice-Fog Formation:
Although many former studies (Thuman and Robinson, 1954; Kumai, 1964;
and Benson, 1965) have suggested that the mechanism of ice-fog formation
is the freezing of water droplets resulting from moisture of car exhaust
or open water, they did not show either verifications or considerations
based upon observations for the purpose. In the present research we have
shown that: 1) Large amounts of steam (small droplets), formed initially
from water evaporated from open water, are formed in the layer very close
to the water surface and disperse into the atmosphere. The evaporation
rate of water vapor from the surface and production rate of water drop-
lets were also estimated, based upon the observations. 2) Aerial photo-
graphs showed steam or water clouds, which are important sources of ice-
fog moisture, coming from the open water of the river, slough and cooling
ponds, power plants and private heating systems. 3) Such water droplets
will freeze in several seconds within a distance of 3 to 5 m from open
water under low temperature conditions. This was confirmed by the
observations of the time required for droplets to freeze, and temperature
profiles above water surfaces. Also, it was supported by calculation of
conductive cooling and radiative cooling, which were also directly measured
by means of a radiometer for various objects in ice fog. Auto exhausts
supplement water droplets and they will be changed to ice-fog crystals in
the same way. So running cars are sprinkling ice crystals rather than
i
adding moisture as vapor into the atmosphere. 4) The humidity in ice/fog
lies between water- and ice-saturation, allowing the ice-fog crystals
132
-------
to grow, but very slowly because of the small differences of saturation
vapor pressure between them. This results in ice-fog crystals having the
smallest size of ice crystals and being suspended in the air for a long
time. The modal size of the ice crystals changes with the changes in the
moisture content of the environmental air. In an area which has less
moisture than ice saturation, the crystal sizes will be smaller and smaller
until the crystals disappear. A typical size distribution of ice fog is
shown in Fig. 6. These observations are also supported by a theoretical
consideration of size distribution of ice-fog crystals. 5) The most
important factor in the form^lon of steam or water droplets is not the
concentration of condensation nude"' in the case of formation of steam from
open water, heating plants and car exhausts under ice-fog conditions but
rather the temperature differences between water (not ice) and ambient
air temperature. However, condensation nuclei or other particles contained
in the water droplets accelerate the freezing of water droplets at a higher
temperature than the spontaneous freezing. We believe that the onset
temperature of ice-fog formation is higher in inhabited areas than in
unpolluted areas due to the greater numbers of condensation nuclei and
effective freezing nuclei in the city, as well as the difference of moisture
I
supply between in-city and out-of-city sites. However, at temperatures
lower than about -37C, the homogeneous nucleation of condensation and
successive spontaneous freezing of water droplets are quite possible even
in contaminated areas. 6) The lowering of air temperature increases the
133
-------
rate of evaporation, i.e. rate of water-droplet formation, speed of the
droplet freezing, suppression of the frozen ice-crystal growth, and form-
ation of more numbers of smaller crystals. Thus, denser ice fog can be
seen at lower temperatures. So the low temperatures are essential for
dense ice fog, providing constant moisture sources are available.
134
-------
9. SYNOPTIC METEOROLOGICAL STUDIES OF ICE FOG
Local radiative cooling as described by Wexler (1936), or as modified
by Gotaas and Benson (1965), has generally been used to explain the low
temperatures and unusually strong inversions observed in interior Alaska.
The association of anticyclones with extreme cold is well known in general,
but little effort has been made to distinguish between the effects of high
pressures produced locally by radiative cooling and those associated with
advection from regions outside Alaska. An attempt has been made in the
present project to show the effects of advective and dynamic processes in
producing, or inhibiting the production of, the low temperatures and to
some extent the steep inversions which lead to the formation of ice fog.
Since the report has already been published (Bowling, Ohtake and Benson,
1968), only the abstract is included here in the section a).
a. Winter Pressure System and Ice Fog in Fairbanks, Alaska:
The production of the low temperatures which are responsible for ice
fog in inhabited areas of interior Alaska would appear to be a classic
example of clear sky radiative cooling under nearly polar night conditions.
However, examination of the meteorological conditions associated with 15
periods of dense ice fog at Fairbanks indicates that local radiative
cooling is important only in producing the observed steep ground inversion.
The most rapid decreases in temperatures at heights > 1 km occurred with
cloud cover and cold air advection preceding the cold weather at the ground.
135
-------
The most common synoptic pattern (observed for the 12 shortest events)
consisted of the migration of a small high from Siberia across Alaska.
Rapid growth of the high was common, and the resulting subsidence was
strong enough to counter-balance not only radiative cooling, but further
cold air advection as well. This resulted in an observed warming aloft
during all but the first 12-24 hr of the clear, cold weather observed at
the ground. Three of the 15 events did not follow this pattern. Two long
and very cold events were associated with warm highs in northeastern
Siberia, continuous belts of moderately high pressure extending from Siberia
across the Bering Strait into Alaska, and advection from Siberia and the
Arctic Ocean. The remaining long but relatively mild event was associated
with a warm high north of Alaska and advection from Canada and the Arctic
Ocean.
b. Analysis of Air Mass Trajectories:
As mentioned previously, water vapor measurements in the vicinity of
Fairbanks show near-ice-saturation in the winter. Since, in general, water
vapor is transported with air mass, it is interesting to study the origin
of dry or wet air masses which arrive at Fairbanks. Using 500 mb synoptic
maps for the winter of 1967 to 1968, an air trajectory analysis was made.
In the analysis, the trajectories we,, j followed for 5 to 7 days before their
arrival in the Fairbanks area, and also at the 500 mb level geostrophic
winds were assumed as predominant. With omission of full trajectory maps
which can be seen in Henmi's (1969) master's thesis, the following results
may be noted: 1) Warm and wet air masses are associated with the
136
-------
trajectories over the Pacific Ocean. 2) Cold and dry air came with air
from the Arctic Ocean, in which air ice fogs occured in the winter of 1968.
3) The cooling of the air in the Fairbanks area was partly caused by the
advection of the cold air mass from the Arctic Ocean. 4) The cessation
of the ice-fog events occurred when air was advected from the Pacific Ocean.
c. Practical Ice Fog Prediction:
Appleman (1953) reported an interesting method of forecasting ice
fogs, which is based upon a combination of temperature and relative humidity.
Although he considered only the effects of burning hydrocarbons under
conditions of high relative humidity, the proposed diagram for the conditions
necessary for formation of ice fog seems to be correct. However, to predict
ice fog using his diagram we have to predict relative humidity as well as
temperature. As previously mentioned, the measurement even of the present
value of humidity is very difficult. Even at the U.S. Weather Bureau,
current relative humidities at temperatures below -35F are reported by
assuming ice saturation and converting to relative humidity over water.
From a) and b) in this chapter we propose the following simple method
for forecasting ice fog. From the middle of November through February
every year, the proposed method may be applicable. If the predicted 500
mb flow has a component from the north over most of central Alaska, ice
fog will occur in the inhabited areas. In practice this type of flow will
occur with a ridge located between the west coast of Alaska and a position
well into Eastern Siberia. The surface expression of the ridge will
137
-------
normally be in central Alaska. (The surface high pressure in itself is
not an adequate predictor, as a warm core anticyclone from the Pacific
may result in high pressure at the ground with warm weather). If a 500
mb ridge is located farther east, so that the upper air at Fairbanks has
a source to the south, relatively warm temperatures aloft will prevent
excessive radiative cooling at the ground. Southwesterly flow is almost
always accompanied by cloud cover, while southerly or southeasterly flow
over the Alaska Range will show a foehn effect, giving clear skies and near
normal temperatures.
As was mentioned before, in interior Alaska during the winter the
humidities are almost ice saturation at the surface even outside the city,
due to the extensive snow cover. This means that we have to watch only
temperature for making ice-fog forecasts, even though Appleman proposed
the method of forecasting considering both temperature and humidity.
However, the present method is sensitive to errors in the prognostic 500
mb charts being made by the U.S. Weather Bureau, especially if a ridge is
located near the Alaskan west coast. During late November, February and
March, it is difficult to predict definite ice fog. In these months ice
fog generally appears only at night. To improve the accuracy of the pre-
diction even for other areas than Alaska the prediction of the planetary
wave around the poles is essential.
138
-------
10. VISUAL RANGE IN ICE FOG
For water fogs and clouds which consist of water droplets, a relation-
ship between the meteorological visual range (= visibility), the liquid
water content and mean diameter of the droplets was originally developed
by Trabert. However, distinctive size distribution and shape of ice-fog
crystals might give a different relation between them.
An article published previously presented the results of an experi-
mental investigation into the relationship between the visual range and
the size distribution of ice-fog crystals at the MUS site, the Eielson
Air Force Base, and the International Airport, Fairbanks. An empirical
function is developed for the constant appearing in the Trabert formula.
Use of this function gives visual ranges that agree with measured values
for size distributions of different width. For a full description, see
the published work (Ohtake and Huffman, 1969).
139
-------
11. ELECTRIC PROPERTIES OF ICE-FOG CRYSTALS
One of the purposes of ice fog research is to find methods of
.eliminating ice fog. In order to provide basic information for feasibility
studies, the electric properties of ice fog crystals have been investigated.
From the viewpoint of general atmospheric electricity, research on the
electric properties of micron-sized ice crystals may lead to new in-
formation about the mechanisms of charge generation which occur in natural
clouds.
A relatively small number of workers has investigated the electric
properties of fog, cloud droplets and snow and ice crystals. Gunn (1955)
—9
reported that warm clouds (20C) carried charges of about 1.3 x 10
_3
coulomb m for particles larger than lOy diameter. Assuming 100 droplets
_3
cm of air, which is obtained from Weickmann and Kampe's (1953) average
concentration of droplets in fair-weather cumulus clouds, from the above
value of the charge of the cloud the individual net charge of droplets is
1.3 x 10 coulomb (81 elementary charges). Twomey (1956) measured the
net charges on individual cloud droplets and reported that the charges in
water clouds were always positive, while negative charges were detected
when ice crystals were present. The negative charges observed on particles
with diameters of 10u were about 1.6 x 10 coulomb (1000 elementary
charges). Scott and Hobbs (1968) measured individual charges acquired by
an ice sphere exposed to snow or cloud particles. These workers found
that individual particles produced charges of both signs ranging from ;
-2-2 77
-10 to +10 esu (from -2 x 10 to + 2 x 10 elementary charges). In
140
-------
the present paper, preliminary results are described for some rather sim-
ple experiments performed on ice fog.
a. Experiments with Ice Fog Crystals in Electric Fields:
i. Ron-Uniform Electric Field:
During the first experiments, ice fog crystals were observed in the
non-uniform field of two parallel copper wires (2.08 mm diameter) mounted
vertically at a distance of 1.5 cm on a bakelite board. Both wires were
coated with enamel and potential differences up to 12 kV could be maintained
between the wires.
The wire electrodes were placed in natural ice fog in Howntown
Fairbanks at temperatures of about -35C. The average concentration of ice
-3
fog crystals was 500 cm . After applying a voltage of 3 kV between the
wires for a period of about 18 hours, we could observe an appreciably
thicker deposit of ice fog crystals on the positive wire than on the neg-
ative one. The deposits occurred in radially oriented dendritic shapes.
For these first observations, the high voltage power supply could
produce only positive voltages with respect to ground. In order to check
why the predominant deposit on the positive wire was affected by the
polarity of the applied voltage, the observations were repeated using
another power supply which was able to produce negative high voltages
with respect to the ground. One minute after applying 12 kV, we again
observed a larger deposit on the positive wire, which was at the same
potential as the ground, than on the negative one. An additional experiment
141
-------
using 3 kV showed the same thing, but the amount of the deposit was less
than that for 12 kV. Unfortunately, since the temperature was as high as
-28C, some particles could have been in the liquid phase (Borovikov, 1968).
Experiments were also performed with artificial ice fog in a cold
chamber which is open at the top only (76 cm x 40 cm x 46 cm high) and
in an environmental cold room (450 cm x 290 cm x 208 cm high). The walls
of both co1d chamber and room are made of stainless steel which can easily
frost on the surfaces. To avoid the possibility of some parts of the x*all
frost flying to the electrodes, the walls of the cold chamber were painted
with enamel and coated with ethylene glycol. On the other hand the walls
of the cold room did not have any coating nor painting.
Also to wipe the deposited ice away from the electrodes, a paper
towel which was soaked with a small amount of glycol was used between
successive experiments. This procedure was applied to eliminate undesir-
able charge generation on the wires and ice in the chambers resulting from
rubbing between ice depositions.
Observations were made every 60 seconds after applying a high voltage.
During the experiment the observer tried to be as far away from the elect-
rodes as long as possible, excepting the times of observation of deposit
on the electrodes. If the observer is standing close to the wires when
the high voltage is applied, he will act as another electrode.
The experiments in the cold chamber were carried out by applying the
high voltages with different polarities; a) 3 kV positive potential/
zero electric potential with respect to the ground, b) 3 kV negative
142
-------
potential and zero potential with respect to the ground, and c) 3 kV
positive and 3 kV negative potential with respect to the ground (potential
between both electrodes, accordingly, was 6 kV) . In the case of a) both
wires received a deposit of ice fog crystals but the amount of deposit on
the positive wire was much more than on the negative wire for the first
60 seconds. 120 seconds after applying the high voltage, the positive wire
had a very much heavier deposit in radially oriented dendritic shapes.
Case b) showed that both wires had approximately equal amounts of deposits
after the first 60 seconds, but the positive wire seemed to have just a
little more deposit than the other wire. After 120 seconds the appearance
was the same as after 60 seconds, but the deposit on the positive wire
developed radially oriented needle shapes, which are an earlier stage of
the dendritic shape. 180 seconds after the voltage was applied, deposit
on both wires developed and the positive wire collected ice crystals in
the shape of radially oriented needles appreciably more than the negative
wire. In c) , also both wires received a deposit, but the positive wire
obviously had more deposit after the first 60 seconds. 120 and 180 seconds
after 6 kV was applied with a neutral ground, much more deposit could be
found on the positive wire than on the negative wire.
In summarizing the experiments using non-uniform fields, the positive
wire always received more deposit than the negative wire. Throughout the
above experiments the temperature in the cold chamber was maintained be-
tween -40C and -45C, and ice fog was produced by evaporation of water
vapor from distilled water (about 5C) in a dish for 30 seconds, & half
minute before applying the high voltage in the chamber. The concentration
143
-------
_3
of ice fog crystals was approximately 2000 cm . Each time, the deposition
of ice crystals on the wires occurred earlier in the chamber than in
natural ice fog outside. This was at least partly an effect of higher
concentration of ice fog crystals. The observer who was standing per-
pendicular to the plane of the two wires could see the development of the
deposition in time, observing that the initial deposition occurred on the
positive wire and that some small parts of the deposition in the shape of
radially oriented needles or dendrites broke off and flew to the other wire
on paths approximately along field lines.
Similar experiments in the environmental cold room showed also the
same results, but they were not very clear compared with those obtained in
the cold chamber. This could be explained by the effects that first some
frost from the walls deposited on the negative wires and second that the
observer tried not to disturb the air and watched both wires from a dis-
tant position. The temperature was kept between -40.OC and -42.5C. Ice
fog was made by mixing the air inside with outside air when the observer
entered in the room. Higher voltages and thicker ice fog caused the ice
crystals to deposit rapidly on the wires.
ii) Uniform Electric Field:
In order to gather further information about the nature of the forces
acting on ice fog crystals in electric fields, ice crystals were observed
in the uniforia field between two parallel plates. The plates were 25 cm
square made of uncoated copper mounted on a wooden board. Separation was
144
-------
1 cm. The edges were rolled with a radius of 4 mm and the corner were
rounded to avoid excessive non-uniform field; (see Fig. 67). The elect-
rodes were placed in natural ice fog parallel to slight winds so that ice
crystals could more easily pass through the field. A voltage of 3 kV
positive and 0 kV with respect to the ground was applied to the plates
and observations were made at temperatures of around -35C.
Even in dense ice fog, no ice particles were deposited on the plane
surface of either electrode during an observation time of approximately 20
hours. Some deposition radially oriented in the directions of electric
field lines was found on the edges of the positive plate where the electric
field is non-uniform.
Artificial ice fog was also observed in a uniform field in both cold
chamber and environmental cold room at temperatures ranging from —47C to
-35C. The uniform field was produced with the same set of electrodes to
which voltages of plus and minus 3 kV with respect to the ground were applied.
No visible deposition of ice-fog particles was observed on the electrodes
in the uniform part of the field during 5 min in much thicker artificial
ice fog than real ice fog. As in the case of natural ice fog, a dendritic
oriented deposit was found on the edges of the positive plate. Even though
high voltages the same as a) and b) for non-uniform fields were applied to
the plates, we found no deposit on the inside surface of the plates in the
uniform part of the field, although many ice-fog crystals were observed in
the region between the plates. The experiment was repeated with electrodes
coated with an insulating layer of varnish. Voltages up to 16 kV with
145
-------
Fig. 67. The parallel plates to produce a uniform electric field.
-------
respect to the wall were applied to the plates, but no deposit of ice fog
crystals was observed on the electrodes in the uniform part of the field.
At the highest electric potential of 16 kV per cm, the uniform field was
appreciably higher than the maximum of the field of two parallel wires
(with a potential difference of 3 kV and a distance of 1.5 cm) which gave
rise to a clearly visible deposit of ice fog crystals on the positive wire
and the negative wire.
b. Discussion:
From the failure of ice-fog crystals to deposit on the electrodes
in a uniform electric field, one might conclude that attractive forces
are sufficiently strong only in non-uniform electric fields to produce
drift motion of the crystals. The equation for the motion of the center
of mass of a particle in an electric field is
MR + KR=*q.. E(r±), CD
where M is the mass of the particle, R is the radius vector to the center
of mass, q is the i-th elementary charge on the particle, E(r^) *s fc^e
electric field at the location of the i-th charge, and K R is the friction
force due to the surrounding medium.
If the electric field is uniform, eq. (1) can be rewritten in the
following form
147
-------
M R + K R - E0I q. . (2)
i
EŁ
o denotes the uniform field. If the net charge Q = . q is zero, it
is apparent from eq. (2) that the total force acting on the particle
vanishes. In this case the motion of the center of mass is unaffected
by the uniform field, i.e., no dritt motion of the ice crystal occurs.
For che motion of a particle in a non-uniform field we must consider
y
the more general eq. (1). The term . q. . ECr^) = F can represent a
non-vanishing force even if Q is zero. This is demonstrated by the
example of a dipole. The sum of the forces exerted on the two opposite
and equal charges is given by the expression
CP • v) E.
where P is the dipole moment. In general, F •, is a non-zero force, in a
non-uniform field. In the case of a uniform field E, however, the dif-
ferential operator on the right hand side of eq. (3) vanishes and Fj is
zero in agreement with eq. (2).
Since no evidence for drift motion of ice-fog crystals in uniform
electric fields has been observed, we can assume that any net charge 0
of tne particles must be insignificantly small. A drift velocity of 0.1
cm sec would produce x^ithin 3 minutes a visible deposit on the attracting
electrode under the ice- fog condition of the cold chamber. As neither aj
depletion of ice fog between the parallel plates bounding the uniform
field nor a deposit has been observed, we can assume that any drift
148
-------
velocity R has been smaller than 0.1 cm sec"1, and an upper bound for the
net charge Q on ice-fog crystals can be derived. From Stokes1 law we
get the relation
n r
— (4)
where ]Ł), r and n are the electric field, the particle radius and the co-
efficient of viscosity, respectively. For r = 5y, [E| = 16 kV cm"1 and
-4 -1 -1
n = 1.510 x 10 g cm sec at a temperature of -40C (Mason, 1957), we
find Q = 6 elementary charges.
The observation of an insignificantly small net charge of ice-fog
crystals is supported by the result of our optical and electron microscope
studies, which do not indicate any effects of generally proposed mechanisms
for charge separation such as freezing iu the presence of thermal gradients,
chemical impurities in small droplets, droplet shattering during freezing
or splintering from larger crystals.
Forces acting on ice-fog crystals only in non-uniform electric fields
can be explained by polarization of the crystals. The force pj exerted
on a dipole moment P is given by eq. (3) and vanishes in a uniform field.
P could be a permanent as well as an induced dipole moment. In either
case P is oriented in the direction of the electric field and Fd has ttie
direction of the field gradient, which points toward the nearest electrode
if one considers only the field in the vicinity of each of the two wires.
The assumption of a dipole moment therefore leads to attractive forces
149
-------
toward both wires, and deposits on both wires have in fact been observed
(see Fig.68a).
An induced dipole moment P. of ice-fog crystals must always be
present, and an approximate value of p. can be obtained from the equation
(Jefimenko, 1966).
p .
1
4 ir r e (e - 1)
e + 2
(5)
where e is the dielectric constant (for ice, it is a tensor quantity) of
the spherically shaped particles of radius r. At a temperature of OC the
components of e along the c-axis and perpendicular to the c-axis are 106
and 92, respectively (Gr'anicher, 1963). Approximate values of P. can be
calculated using the scalar dielectric constant in eq. (5) as the value
e = 99 and the field strength Ł of the non-uniform field. The potential
ij; of two parallel cylinders of the same radius p, whose center lines are
oriented along the z-axis and intersect the x-axis at +d and -d, is given
by the following equation:
log!
2 log
[(x-a) +
[(x+a) + 7 f
(6)
where (j> is the potential difference of the cylinders, and a has the value
of
2 2 *
a = (dZ - p^) (7)
150
-------
a.
b.
c.
Fig. 68. Drawings for the possible explanations of ice-fog crystal deposi-
tions on the parallel wires. The upper drawing (a) shows ice crystals
which have induced dipole moments; (b) shows crystal which has dipole
layer on the surface; and (c) shows forces ^exerted to the ice crystals
near positive and negative electrodes.
-------
Using <|> = 6kV, p = 1.04iran and d = 8.5 ram, which correspond to the
experimental conditions, one can compute the electric field Ł at the inside
wire surface (x = 7.46 mm, y = 0 mm) from eq. (6), and from eq. (5) one gets
— 9 1
for the magnitude of the induced dipole moment P. the value 1.58 x 10~
amp sec ra. This value is based on a particle radius r = 5y. The field
9 -2
gradient at the same location is 1.14 x 10 V m , and the resulting force
I pjj on an ice-fog crystal is 1.8 x 10 dyne. For a viscosity
-4 -1 -1
n = 1.510 x 10 g cm sec , this force corresponds to a drift velocity
I* t —1
R = 0.13 cm sec . Although the force p due to an induced dipole moment
' d
in non-uniform field is sufficiently large to produce significant drift
motion of ice-fog crystals, the existence of a permanent dipole moment
cannot be ruled out on the basis of the experiments reported in this paper.
The observation that ice crystals are preferentially deposited on
the positive wire cannot be explained by a dipole moment of the particles.
A possible explanation can be seen in Weyl's (1951) model of ice crystals.
Weyl postulated that, due to the crystal structure at the surface, ice
crystals are covered by a dipole layer such that the extreme outer layer
in negative (Fig. 68b). A schematic drawing of Weyl's model is shown in
Fig. 2. The dipole moment p per unit area is directed toward the interior
of the particle. The force exerted on a crystal is
E ds C8)
152
-------
S is the crystal surface and ds denotes the surface element. Con-
sidering that the dominating contribution to the integral comes from
surface elements closest to a wire, we can see that the resulting force
attracts particles toward the positive electrode and repels them from
the negative wire thus leading to a predominant deposit on the positive
electrode (Fig. 68c). From the discussion of equations (1) and (2) it
follows that F-, vanished in a uniform field. Further investigation of
p could be of interest to the physics of ice crystal surfaces and related
phenomena.
153
-------
REFERENCES
Appleman, H., 1953: The cause and forecasting of ice fogs, Bull. Amer.
Met. Soc.. 34, 397-400.
Benson, C. S., 19.65: Ice fog: Low temperature air pollution defined
with Fairbanks, Alaska as type locality. Geophysical Institute
Rept.. UAG R-173.
Blanchard, D. C., 1951: A verification of the Bally-Dorsey theory of
spicule formation on sleet pellets, Journ. Meteor., 8, 268-269.
Borovikov, A. M., 1968: Supercooling of water in the atmosphere and
the phase of various type of clouds. Proc. Intern11. Conf. on
Cloud Physics, Aug. 1968, Toronto, Canada, 290-294.
Bowling, S., T. Ohtake and C. S. Benson, 1968: Winter pressure systems
and ice fog in Fairbanks, Alaska. J^ App^. Met., 7, 961-968.
Branton, C. I., 1965: A proposed technique for measuring relative
humidity at below freezing temperature, Humidity and Moisture,
New York, Reinhold Pub. Corp., 95-100.
Byers, H. R., 1965: Elements of cloud physics, Chicago, Univ. of Chicago
Press, pp. 191.
Dorsey, N. E., 1948: The freezing of super cooled water, Transactions
Ame. Phil. Soc., 38, 247-328.
Fletcher, N. H., 1966: The physics of rainclouds, Cambridge,
Cambridge University Press, pp. 390.
Fournier d'Albe, E. M., 1949: Some experiments on the condensation
of water vapor at temperature below 0°C, Quart. Jour. Roy. Met^.
Soc.. 75, 1-14.
Gotaas, Y., and C. S. Benson, 1965: The effect of suspended ice crystals
on radiative cooling, J_. Appl. Met., 4, 446-453.
Granicher, H., 1963: Properties and lattice imperfections of ice crystals
and the behavier of H20-HF solid solution, J_. Physik kondensierter
Materie, Vol 1, 1-12.
Gunn, R., 1955: Droplet-electrification processes and coagulation in
stable and unstable clouds, J. Meteor., 12, 511-518.
154
-------
Hanajima, M., 1944: Supplement of artificial snow-crystal growth, J_.
Met. Soc. Japan, Ser. II, 22, 121-127.
Henmi, T., 1969: Some physical phenomena associated with ice fog,
Master's thesis at the Univ. of Alaska, pp. 90.
Hobbs, P. V., 1965: The aggregation of ice particles in clouds and
fogs at low temperatures, J_. Atmos. Sci.. 22, 296-300.
Houghton, H. G., 1951: On the physics of clouds and precipitation,
Compendium of Meteorology, Boston, Amer. Met. Soc., 165-181.
Huffman, P. J., 1968: Size distribution of ice fog particles,
Master's thesis at the Univ. of Alaska, pp. 93.
Huffman, P. J., and T. Ohtake, 1970: Formation and growth of ice fog
particles at Fairbanks, Alaska, submitted to J_. Geophys. Res.
Isono,- K., 1969: On the term "Hyoshokaku (ice nucleus)", Tenki,
16, 83-84.
Jefimenko, 0. D., 1966: Electricity and Magnetism, Applets-Century-
Crofts, New York, Division of Meredith Publishing Co., 591.
Junge, C., 1953: Die Rolle der Aerosole und der gasfb'rmigen
Beimengungen der Luft im Spurenstoffhaushalt der Troposphere,
jTellus, 5, 1-26,
Kline, D. B. and G. W. Brier, 1961: Some experiments on the measurement
of natural ice nuclei, Monthly Weather Review, 89, 263-272.
Kobayashi, J., 1960: Investigation on Hygrometry, Papers in Meteorology
and Geophys., 11, 213-338.
Kobayashi, T., 1956: Experimental researches on the snow crystal habit
and growth by means of a diffusion cloud chamber (preliminary
report), Low Temperature Sci.. 15, 1-12.
Kobayashi, T., 1965: Vapor growth of ice crystal between -40 and
-90°C, J_. Met. Soc. Japan, Ser. II, 43, 359-367.
Kocmond, W. C., 1965: Investigation of warm fog properties and fog
modification concepts, 2nd Annual Report, CAL Report No. RM-
1788-P-19.
Kohler, H. 1925: Untersuchungen uber die Elemente der Nebels und
der Wolken, Meddel. Met.—Hydr. Aust. Stockholm, 2, No. 5.
155
-------
Kozima, K., T. Ono and K. Yamaji, 1953: The size distribution of
fog particles, Studies on Fogs, 303-310, Sapporo, Tanne
Publishing Co.
Kumai, M. 1964: A study of ice fog and ice-fog nuclei at Fairbanks,
Alaska, Part 1, CRREL Res. Rep. 150, pp. 27.
Kumai, M. 1951: Electron-Microscope study of snow crystal nuclei
J.. Met.. 8, 151-156.
Kumai, M. and H. W. O'Brien, 1964: Method of measuring of total
water content by absorbing agent, CRREL Technical report (informal).
Kumai, M. and H. W. O'Brien, 1965: A study of ice fog and ice-fog nuclei
at Fairbanks, Alaska, Part II, CRREL Res. Rep. 150. 14 pp.
Kumai, M. and K. E. Francis, 1962: Nuclei in snow and ice crystals on
Greenland ice cap under natural and artificially stimulated
conditions, _J. Atmos. Sci.. 19, 474-481.
Kuroiwa, D., 1951: Electron microscope study of fog nuclei,
,J. Met., 8, 157-160.
Landsberg, H., 1938: Atmospheric condensation nuclei, Ergebn. Kosm.
Phys., 3, 155, [or see Fletcher, 1966],
Langer, G., J. Rosinski and C. P. Edwards, 1967: A continuous ice
nuclei counter and its application to tracking in the troposphere,
J. Appl. Met.. 6, 114-125.
Mason, B. J., 1957: The physics of clouds, Oxford, The Clarendon Press,
481 pp.
Munn, R. E., 1966: Descriptive micrometeorology, New York and London,
Academic Press, 245 pp.
Nakaya U., 1954: Snow crystals, Cambridge, Harvard University Press,
510 pp.
Ogiwara, S. and T. Okita, 1952: Electron microscope study of cloud
and fog nuclei, Tellus, 4, 233-240.
Ohtake, T., 1964: An airborne cloud-droplets sampler, Sci. Rep. Tohoku
Univ., Ser. 5, Geophys. 15, 59-65.
Ohtake, T., 1967: Alaskan ice fog. Phys. of Snow and Ice, Part I,
Hokkaido Univ., Sapporo, 105-118.
Ohtake, T. and P. J. Huffman 1969: Visual Range in Ice Fog, Ł. Appl.
Met. 8, 499-501.
156
-------
Ohtake, T. and H. Isaka, 1964: Determination of effectiveness of
artificial stimulation snow in Tohoku District, Japan, SCJL.
* Tohoku Univ.. Ser. 5, Geophys., 15, 97-110.
Okita, T., 1958: Water blue film method for measurement of cloud and
fog droplets, J_. Met. Soc. Japan, 36, 164-165.
•Oliver, V. J. and M. B. Oliver, 1949: Ice fog in the interior of
Alaska, Bull. Amer. Met. Soc., 30, 23-26.
Radke, L. F. and P. V. Hobbs, 1969: An automatic cloud condensation
nuclei counter, _J. Appl. Met . , 8, 105-109.
Robinson, E., G. B. Bell, W. C. Thuman, G. A. John and E. J. Wiggins,
1954: An investigation of the ice fog phenomena in the Alaskan
area, Final Report, Contract No. AF19(122)-634, Stanford Res. Inst .
Schaefer, V. J., 1962: The vapor method for making replicas of liquid
and solid aerosols, J_. Appl. Met . , 1, 413-418.
Scott, W. D. and P. V. Hobbs, 1968: The spectra of changing events
due to the collision of natural ice particles with an ice surface,
Quart . JL Roy. Met.. Soc. , 94, 510-522.
Shimizu, H., 1963: "Long prism" crystals observed in precipitation
in Antarctica, J_. Met. Soc. Japan Ser. II, 41, 305-307.
Stine, S. L., 1965: Carbon humidity elements — Manufacture, performance
and theory, Humidity and Moisture, New York, Reinhold Pub, Corp.,
316-330.
Stoll, A. M. and J. D. Hardy, 1955: Thermal radiation measurements
in summer and winter Alaskan climates, Trans . Ame. Geophys . Union,
36, 213-226,
Tanaka, T. and K. Isono, 1966: A technique for identification of
substances of ice nuclei, (in Japanese). Preprint of Japan. Met.
Soc. Meeting May, 1966.
Thuman, W. C. and E. Robinson, 1954a: Studies of Alaskan ice-fog
particles, J_. Met. 11, 151-156.
Thuman, W. C. and E. Robinson, 1954b: A technique for the determination
of water in air at temperatures below freezing, J_. Met. , 11, 214-219.
Twomey, S., 1956: The electrification of individual cloud droplets,
Tellus, 8, 445-452.
157
-------
Tyndall, A. M., 1922: Proceedings of Physical Society, (London), 34,
72.
Warner, J., 1957: An instrument for the measurement of freezing
nucleus concentration, Bull. Obs. Puy de Dome, No. 2, 33-46.
Weicfcmann, H., 1948: Die Eisphase in der Atmosph'are, Ber. deutch.
Wetterd. U. j>. Zone, No. 6, 54 pp.
Weickman, H. K. and H. J. aufra Kampe, 1953: Physical properties of
cumulus clouds, J_. Meteor., 10, 204-211.
Weller, G., 1969: Ice fog studies in Alaska, Geophysical Institute
Report, University of Alaska, UAG R-207, 49 pp.
Wexler, H., 1936: Cooling in the lower atmosphere and the structure
of polar continental air, Mon. Weather Rev., 64, 122-136.
Weyle, W. L., 1951: Surface structure of water and some of its
physical and chemical manifestations, J^. Coll. Sci., 6, 389-405.
Yamamoto, G. and T. Ohtake, 1953: Electron microscope study of cloud
and fog nuclei, Sci. Rep. Tohoku Univ. Ser. 5, Geophys. 5,
141-159.
Yamamoto, G. and T. Ohtake, 1955: Electron microscope study of cloud
and fog nuclei II, Sci. Rep. Tohoku Univ. Ser. 5, Geophys. 7,
10-16.
158
-------
APPENDIX
THEORETICAL STUDY OF ICE-FOG SIZE DISTRIBUTION
by
P. J. Huffman
AFCRL, Bedford, Massachusetts
and
T. Ohtake
Geophysical Institute, University of Alaska, College, Alaska
159
-------
ABSTRACT
A mechanism is proposed for the formation of ice fog particles in
the city and environs of Fairbanks, Alaska. Equations are developed for
calculating the size distribution resulting from growth by sublimation
of water vapor. These equations are solved numerically, with the use of
a computer, for three major types of ice fog sources: A) automobile
exhaust; B) exhaust from heating plants and C) open water. For source
type A, the computed size is much smaller than that observed; but for
source types B and C, the computed size distributions are found to be
in good agreement with experimentally obtained values.
The discrepancy between the computed and observed sizes for source
type A is possibly due to the large degree of supercooling exhibited by
small water droplets. Only very few of the small water droplets, initially
formed by condensation, freeze. These frozen particles may then grow to
the observed size at the expense of the more numerous evaporating liquid
droplets. On the other hand, the larger droplets produced by source types
B and C probably nearly all freeze very shortly after their formation,
leaving no appreciable supply of liquid droplets to provide for their
further growth.
160
-------
1. INTRODUCTION
In an area as sparsely populated as interior Alaska, air pollution
at first sight would not appear to be a problem. During periods of clear
sky, the long winter nights of the arctic permit extreme radiative cooling
of the earth's surface. In protected valley sites, this radiative cooling
is often responsible for strong temperature inversions of extended duration
(Benson, 1965). One such location is the city and environs of Fairbanks.
At temperatures below -30C, a large concentration of microscopic ice
fog particles* is normally present in the inversion layer. The most
prominent feature of the ice fog is a severe restriction in visibility
through the lowest part of the atmosphere. At present, ice fog is most
pronounced in and around Fairbanks city, but it is a major problem which
must be considered in the development of interior Alaska or the arctic
regions in general.
The first detailed study of ice fog particles was made by Thuman and
Robinson (1954). They found that the mean statistical diameter decreased
with decreasing temperature. At -40C, the mean diameter of the irregular
shaped particles, referred to as droxtals, was 13y. Kumai (1964) found
The term "ice fog particle" is used throughout to describe the elemental
solid objects, regardless of shape, that compose the ice fog phenomenon.
The purpose of this nomenclature is to distinguish ice fog particles from
"ice crystals", the latter term being reserved to refer to the products
of sublimation and freezing occurring in the higher atmosphere. Ice fog
particles may be symmetrical, but are more often irregular in shape. The
term "ice fog particle" does not imply a lack of crystal structure but
refers rather to the mechanism of production.
161
-------
that most ice-fog particles were between 2y and 15p in size with a sharp
peak in the distribution near 7y. More recent measurements (Huffman,
1968) show that the size distribution varies with location and often has
more than one mode. Electron microscope analysis (Ohtake, 1967) reveals
that the nuclei of most ice-fog particles are located well away from the
geometric center and many ice-fog particles have no apparent nucleus at all.
Recent measurements by Henmi (1969) using the method of Ohtake (1968), show
that during ice fog conditions the ambient atmospheric water vapor content
lies between saturation values with respect to water and ice.
Considerable theoretical attention has been given to the nucleation
and growth of ice crystals in clouds (Mason, 1957; Fletcher, 1962; Byers,
1965). This paper deals with the theoretical treatment of the nucleation
and subsequent growth of ice-fog particles. In the upper atmosphere,
water droplets and ice crystals form when the air cools gradually. Through-
out the growth period, the air mass is not far removed from the saturation
level. On the other hand, it will be shown that ice-fog particles are
produced by the injection of saturated warm water vapor directly into a
cold environment. The condensation and subsequent freezing takes place
at a higher degree of supersaturation and at much greater cooling rates
than in the former case. As the ice-fog particles diffuse away from
their source of production, any remaining water droplets soon evaporate,
providing for the further growth of existing ice-fog particles by sublima-
tion and maintaining the environment near ice saturation.
162
-------
2. THEORETICAL CONSIDERATIONS
Consider an exhaust gas containing water vapor at initial temperature
T. injected into an environment at temperature T , T. » T . The saturation
•*• o i o
ratio first increases as the gas begins to cool and then decreases as con-
densation proceeds. At any time, the saturation ratio S is given by
S = J- , (1)
s
where e is the water vapor partial pressure and e is the saturation vapor
S
pressure at the same temperature, The time rate of change of the saturation
ratio is thus
dS = 1_ de S_ fS m
dt e dt ~ e dt
s s
Using the ideal gas law for water vapor, the first term of (2) can be
rewritten as
3^_ jle RT pv S^ dT_
ee <*t e M dt~ + T dt ' *• '
S o
where R is the gas constant, T is the absolute temperature, M is the
molecular weight of water and p is the water vapor density. The time
rate of change of water vapor density is
dp dm.
dT = "] dT • (4)
where m is the particle mass, the summation being over all particles per
unit volume. Substituting (4), (3) can be expressed as
163
-------
_
e dt T dt e M * dt
8 8 J
The rate of change of saturation vapor pressure with time can be
expressed as
_s s dT . dT
dt = dT dt ^p~ dt
where the Clausius - Clapeyron equation for an ideal gas has been used.
In (6), L is the latent heat of the phase change.
Inserting (5) and (6) into (2) gives
IM dT RT
_
v '
dt T RT' dt e M dt
s J
For the temperatures with which we are concerned (23 3K j? T <_ 333K) , LM/RT
is always considerably greater than unity and (7) may be simplified to
dS _ LMS dT RT_ f*i f .
dt RT2" dt ~ egM ^ dt w
The appropriate values for e and L in (8) are the average values for the
S
system of particles. Condensation first occurs to form water droplets
which then freeze and continue to grow by sublimation. In general, the
system consists of both water droplets and ice particles. It is assumed
that the probability of freezing P for a droplet of radius r, supercooled
to a temperature T after t seconds is given by* (Bigg, 1953)
The validity of (9) for ice fog is doubtful since the referenced literature
deals with much lower cooling rates than those encountered here. The absolute
temperature, T, is that measured in the environment surrounding the particles.
This is essentially the same as the particle temperature only for sufficiently
small cooling rates. However, since the values of eg and L for ice do not
differ greatly from the corresponding values for water, use of (9) for
computing these paramenters does not introduce appreciable error.
164
-------
In (1-p) m - Ar3t (exp T - 1), (9)
s
0 ; T > Tffi
- T ; T
where A = 6.5 x 10" cm"3 sec"1 and Tm = 273°K is the melting temperature.
Thus es and L have the form
es = BW + {1 - exp [-Ar3t (eTs - 1)]} (ejL - ew) (lOa)
L = Ly + {1 - exp [-Ar3t (eTs - 1)]} (Ls - LV) (lOb)
where e± and ew are respectively the saturation vapor pressures over ice
and water; and L^ and Ls are respectively the heats of vaporization and
sublimation.
Three major types of ice fog sources can be classified: A) automobile
exhaust; B) exhaust from heating plants (commercial and residential) and
C) open water. Benson (1965) has shown that the rate of cooling of
automobile exhaust along the centerline of the exhaust plume is given by
the empirical relationship.
4 = -a (T - T)2; T > T (11)
where x is the distance from the source and a is a constant. Air temperature
measurements (Ohtake, 1968) indicate that the temperature gradient above open
water can also be expressed by (11); the temperature drops from a value near
165
-------
OC at the water surface to the ambient value at a height of about 2 meters.
Thus, the theoretical aspects of ice fog formation from exhaust sources and
from sources of open water can be traced in the same mathematical manner,
subject of course to different boundary conditions.
Integration of (11) gives the temperature as a function of distance
from the source by
T = TI ~ T° + T0 (12)
(T, - T ) ax + 1
i o
Near the source, a reasonable expression for the velocity of the exhaust
gas as a function of distance is
v = v0 e~x/b , (13)
where vo is the initial velocity and b is a constant, the value of which
depends on the geometry of the source. Integrating (13) and substituting
into (12) gives the temperature of an elemental volume of exhaust gas as
a function of time by
ab(T± - T0) ln(ct
+ TQ , (14)
where c = vo/b. Figure 1 shows the temperature as a function of time
given by (14) for typical values of a, b, T^ and TQ.
Neglecting curvature and chemical effects, the diffusion growth
rate of water droplets and ice crystals is usually given by the classical
expression
LM RT
- , v = j-^
KRT2 esu
166
-------
where C is the capacitance of the particle in air, K is the thermal con-
ductivity of air and D is the diffusivity of water vapor in air. Equation
(15) is not strictly correct when the growth rate is appreciable. If it
is assumed, for mathematical simplicity, that the particle is approximately
spherical throughout the growth period, the growth rate can be expressed
more correctly by (Rooth, 1957)
where f is a correction terra that takes into account the kinetics of
vapor molecules and the accomodation coefficient of the particle surface.
The exact value of f as a function of temperature is not known for water
or ice because information regarding the variation of the accommodation
coefficient is incomplete. We chose the value f = 5 v based on the
measured value of the condensation coefficient at IOC and 100 mb (Alty
and Mackay, 1935). Using (16), the summation in (8) can be expressed as
ft 2
(S-l) I(t ) r ^n* fc) dt ,17v
Jo n f+r (tn> t) n (17)
where I(t ) is the rate of formation of embryo droplets per unit volume
at time t . Thus r (tn, t) is the radius at time t of particles formed
at time tn.
The cooling rates encountered (Fig. 1) are sufficiently great to
produce the degree of supersaturation required for homogeneous nuclea-
tion. Electron microscopy also reveals that ice-fog particles are
possibly nucleated homogeneously since some of the particles did not
167
-------
UJ
-40 -
10'
10
I0~2 IO"1 10°
TIME (SECONDS)
10
10'
Fig. 1. Typical examples of temperature of exhaust gases computed from equation (14).
Curve A: automobile exhaust, a = 5 x 10~4 cm'1 deg"1, b = 66. 7 cm'1, VQ = 2000 cm
sec"1. -Curve B: exhaust from heating plant, a = 2 x 10~4 cm"1 deg"1, b = 667 cm'1,
v0 = 200 cm sec . Curve C: above open water, a = 5 x 10~4
cm
-1
v0 = 20 cm sec
-1
cm " deg" ,
b = 6670
-------
contain larger nuclei in their replicas (Ohtake, 1967). In the use
of the replica method, it is normally difficult to produce very clean
supporting films. Even though we have tried hard to make cleaner films
for this purpose, so far most ice fog particle replicas contained very
small nuclei or contaminations. However, many ice fog particles taken at
Chena Hot Springs, where there is a much smaller amount of air pollution,
did not have any nuclei or contamination in the replicas. Also, it is
possible that the aerosol could have been captured by an existing drop-
let previously nucleated homogeneously, In this paper it is assumed
that ice fog particles are nucleated homogeneously and the presence
of any foreign matter has no effect on the nucleation process. The
homogeneous rate of embryo formation is given by (Farley, 1952).
e2S2 _ 3W 1/2 -, 23
T s /2nMo. , 16-irnM a -, *-„,
I = -2~2 (— ) exp { 332 > , (18)
RTT * 3RVln S
where a is the surface tension of water and n is Avagrado's number.
The set of Eqs. (8), (14), (17) and (18), with eg and L determined
by Eqs. (lOa) and (10b), can be solved numerically to give the resulting
size distribution as a function of the source parameters and the ambient
temperature.
3. COMPARISON WITH EXPERIMENT
A computer was programmed to calculate the saturation ratio as a
function of time. The results of this calulation are shown in Fig. 2
for the three cooling rates of Fig. 1. From the time dependence of S,
the size distribution was evaluated by calculating l(tn) versus r(tn, t)
169
-------
A' IO"4
B' IO"2
C« IO"1
IO"3
IO"1
10°
10-2
10°
IO1
9
8
7
6
€0
T 1
I I
1 I I I I I
TIME (SECONDS)
Fig. 2. Saturation ratio versus time for the cooling rates of Fig. 1.
-------
in the limit as t becomes large without bound. Figure 3 shovrs the
size distribution obtained in this manner.
It would be impossible for ice-fog particles represented by curve
A of Fig. 3 to exist very long after the saturation ratio begins to
diminish; such small particles would evaporate almost immediately after
diffusing from the source. It must be remembered, however, that only
growth by diffusion has been considered. The 0.05 p peak could be
shifted toward a larger diameter by coalescence. Calculations show
that the maximum rate of droplet production is of the order of 10
cm~3 sec~l. Even with such a high concentration of embryo droplets
and assuming a coalescence efficiency of one hundred percent (a factor
difficult to justify), the collision frequency between droplets due to
Brownian motion would not be sufficient for appreciable growth by
coalescence during the short time period available.
Because of the small size of the particles represented by curve
A of Fig. 3, probably only a few of them freeze; these may then grow
to a diameter of several micrometers at the expense of the more
numerous, evaporating liquid droplets. This important effect has
not been taken into account by the computational procedure which only
permits calculation of the average growth rate for the system consist-
ing of both liquid droplets and frozen particles. The frozen particles
grow faster than the liquid droplets because the saturation vapor
pressure is lower for ice than for water. In fact, there comes a time
in the growth period when the saturation ratio falls below unity for
171
-------
o
<
UJ
o
o
o
UJ
UJ
QC
.05
.10
.15
.20
.25
05 10 15 20 25
PARTICLE DIAMETER (MICRONS)
Fig. 3. Computed size distributions for ice fog particles produced
by the cooling rates shown in Fig. 1.
-------
the small liquid droplets and they evaporate, liherating additional
water vapor for further growth of the frozen particles. If the system
consists predominantly of liquid droplets, the few existing ice fog
particles can grow appreciably by this process. On the other hand,
the larger particles represented by curves B and C of Fig. 3 probably
nearly all freeze very soon after their formation and thus, in con-
trast to the previous case, there is not a large amount of excess
water vapor available from liquid droplets to provide for rapid
growth by sublimation.
Figure A*1 presents typical experimental ice-fog particle size
*2
distributions obtained by impaction at several locations (see Fig. 5)
in the vicinity of Fairbanks, Alaska at temperatures between -32C and
-40C (Huffman, 1968). Location 1 (MUS) is in the downtown area. The
Chena River adjacent to location 2 (IAP) is always free of an ice
cover due to the operation of an electric power plant slightly upstream.
Locations 3 (AIRPORT) and 4 (EIELSON AFB) are near airport facilities.
The shape of the distribution at each location is usually (but not
always) as shown in Fig. 4,*^ although the position of the peaks may
be shifted by one or two microns.
The experimental results compare favorably with the computed
size distributions if we assume curve A of Fig. 3 to be shifted
several microns as previously discussed. The usual occurrence of a
1. Fig. 4 is omitted from this appendix, same as Fig. 3 of the text.
2. Fig. 5 is omitted from this appendix, same as Fig. 2 of the text.
173
-------
trimodal distribution at location 1 (MUS) in the downtown area with
peaks near 3.5, 6 and 12 y diameter is believed due to the prevalence
of the three types of ice-fog sources: automobile exhaust, exhaust
from heating plants, and open water respectively. At location 2
(TAP) the distribution is dominated by a single broad peak near 10 p
diameter. We expect this peak to be that of ice-fog particles produced
by open water of the Chena River. The occurrence of a single narrow
peak of 3.5 v diameter at locations 3 and 4 is attributed to automobile
and aircraft exhaust.
4. CONCLUSION
The size of ice-fog particles can be derived on the basis of
diffusion growth with the assumption of homogeneous nucleation if the
source temperature is not too high (i.e. open water and exhaust from
heating plants for which T.^ - 30C) . Because of the nature of the
interdependence between the instantaneous size distribution, satura-
tion ratio and growth rate, the method of computation must be numerical
rather than analytical.
For higher temperature exhaust sources (i.e. automobile exhaust
for which T^ - 60C), the computational method results in ice-fog
particle sizes about 60 times smaller than those observed. Because of
the small size of the. droplets initially produced by such sources, only
very few of them probably freeze and grow to sizes of several micrometers
by sublimation at the expense of the more numerous evaporating water
droplets. This important effect, which is not taken into consideration
by the computational procedure, may explain the reason for the
174
-------
discrepancy between the derived and observed size distributions for
higher temperature exhaust sources. Such sublimation growth for
lower temperature sources is probably negligible since the larger
sized particles produced by such sources should nearly all freeze
very shortly after formation, leaving no appreciable concentration
of water droplets from which to grow further.
The most likely cause of the discrepancy between the observed
and computed size distributions other than that discussed above is an
incorrect assumption for the source velocity, Eq. (13) or somewhat
incorrect choices for one or more of the values for the parameters a,
t>» v0, TŁ. Since the values for these parameters vary considerably
among sources of the same type, and since at present the experimental
data available is limited, the values chosen may not be typical.
Perhaps further comment on the nucleation of ice fog particles is
in order. It was assumed that nucleation occurs homogeneously from
the vapor to the liquid phase. Certainly much particulate matter
is ejected by exhaust sources along with the water vapor; and much
of this particulate matter is capable of serving as condensation
nuclei. A thorough study of the nucleation process must include de-
tailed information of the active heterogeneous nuclei concentration
near the source as a function of temperature and saturation ratio.
However, if no heterogeneous nuclei were present, the cooling rates
encountered are sufficient to cause the degree of supersaturation
required for homogeneous nucleation and therefore, it is water vapor
175
-------
that is of primary importance in the formation of ice fog. In fact,
the cooling rates are so rapid that even the presence of large con-
centrations of heterogeneous nuclei should not be expected to quench
the rapidly rising supersaturation before the critical value for homo-
geneous nucleation is reached. In general then, both homogeneous and
heterogeneous nucleation should be effective.
ACKNOWLEDGEMENTS
The programming required for the numerical computations was
performed by Mr. T. Spuria of Analysis and Computer Systems, Inc.,
Burlington, Massachusetts. This research was supported in part by
National Center for Air Pollution Control, Department of Health,
Education and Welfare, Public Health Service, under grant AP-00449.
176
-------
REFERENCES
Alty, T. and C. A. Mackay, 1935: The accommodation coefficient and the
evaporation coefficient of water. Proc. Roy. Soc., A, 199, 104-116.
Benson, C. S., 1965: Ice fog: low temerature air pollution. Geophysical
Institute Report UAG R-173, 43, (DDC No. 631553).
Bigg, E. K., 1953: The supercooling of water. Proc. Phys. Soc., B, 66,
688-694.
Byers, H. R., 1965: Elements of Cloud Physics. Chicago, University of
Chicago Press, 191 pp.
Farley, F. J. M., 1952: The theory of the condensation of super-
saturated ion-free vapor. Proc. Roy. Soc., A, 212, 530-542.
Fletcher, N. H., 1962: The Physics of Rain Clouds. Cambridge,
Cambridge University Press, 386 pp.
Henmi, T., 1969: Some physical phenomena associated with ice fog.
Unpublished M. S. Thesis presented to the Faculty of the
University of Alaska, 90 pp.
Huffman, P. J., 1968: Size distribution of ice fog particles.
Unpublished M. S. Thesis presented to the Faculty of the
University of Alaska, 93.
Kumai, M., 1964: A study of ice fog and ice fog nuclei at Fairbanks,
Alaska. CRREL Res. Rept. 150, part I (DDC AD451667).
Mason, B. J., 1957: The Physics oŁ Clouds. Oxford, Clarendon
Press, 481.
Ohtake, T., 1967: Alaskan ice fog. Phys. of Snow and Ice, Part 1,
Sapporo, Hokkaido University, 105-118.
Ohtake, T., 1968: Freezing of water droplets and ice fog phenomena.
Proc. Int. Conf. Cloud Phys.. Toronto, 285-289.
Rooth, C., 1957: On a special aspect of the condensation process and
its importance in the treatment of cloud particle growth. Tellus,
9, 372-377.
Thuman, W. C. and E. Robinson, 1954: Studies of Alaskan ice-fog
particles. Ł. Meteor., 11, 151-156.
177
-------
UNCLASSIFIED
DOCUMENT CONTROL DATA - R & D
Geophysical Institute
University of Alaska, College, Alaska 99701
2*. REPORT SECURITY CLASSIFICATION
UNCLASSIFIED
21>. GROUP
3. REPORT TITLE
Studies on Ice Fog
Security Clor.Mfn-Blion
4. DESCRIPTIVE NOTES (Typo ol report and Inclusive dele*)
Final Report - June 1970
3. AUTHOFUSI (First nomo. middle initial, last name)
Takeshi Ohtake
6. REPORT DATE
June 1970
80. CONTRACT OR GRANT NO.
AP-004^9
o. PROJECT NO.
c.
d.
"la. TOTAL NO. OF PAGES
177
'/b. NO. OF REFS
67
9o. ORIGINATOR'S REPORT NUM8ERIS)
UAG R-211
9b. OTHER REPORT NO(S) (Any other numbers that may be *sstŁne3 j
this roper I) I
10. DISTRIBUTION STATEMENT
Distribution Unlimited
It. SUPPLEMENTARY NOTES
12. SPONSORING MILITARY ACTIVITY
National Center for Air Pollution Control
Department of Health, Education Ł Welfare
In order to clarify the mechanism of ice-fog formation, various atmospheric
factors in ice fogs such as size and concentration of ice fog crystals, condensation
nuclei and ice nuclei, amount of water vapor, temperature profile near the sources of
ice fog, etc. were measured.
Nuclei of the ice-fog crystals were studied by use of an electron microscope and
electron-diffraction. The examination showed that most nuclei of ice fog crystals
were combustion by-products and many individual crystals collected near open water
did not have a nucleus, especially at temperatures below -40C. Dust particles or
particles from air pollution are not essential for formation of ice fog; they merely
stimulate freezing of water droplet? at higher temperatures than the spontaneous
freezing temperature. The essential factor is to first form many water droplets
in the atmosphere through condensation of water vapor.
Based on these measurements and calculations of time required for water droplets to
freeze, a physical mechanism of ice fog formation is proposed as follows: 1) Water
vapor coming from open water which is exposed to a low temperature atmosphere, plus
water vapor.from various exhausts of combustion processes is released into the almost
ice-saturated atmosphere and condenses into water droplets, 2) The droplets freeze
very shortly after their formation and before entirely evaporating, 3) Such ice
particles do not evaporate or grow much and stay in the atmosphere with insignificant
fall out, and 4) These processes operate more efficiently in colder environments, which
make ice fog more serious at lower temperatures.
DD.TvV.473
UNCLASSIFIED
Security Classification
-------
JiNjrj A^i F| Fn.
Security ClnBBHM .'lion
KEY WORDS
ROLE WT
ROLE WT
ROLE WT
Condensation
Ice Nuclei
Ice Fog Crystals
Precipitation
Formation
Radiation Cooling
UNCLASSIFIED
Security Classification
------- |