UlTt9 1978
ewiarcti ,-ird Development
Study of the
Subarctic Heat
Island at
Fairbanks, Alaska
L;'J- /
/
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:
1. Environmental Health Effects Research
2 Environmental Protection Technology
3. Ecological Research
4 Environmental Monitoring
5 Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7 Interagency Energy-Environment Research and Development
8. "Special" Reports
9 Miscellaneous Reports
This report has been assigned to the ENVIRONMENTAL MONITORING series.
This series describes research conducted to develop new or improved methods
and instrumentation for the identification and quantification of environmental
pollutants at the lowest conceivably significant concentrations. It also includes
studies to determine the ambient concentrations of pollutants in the environment
and/or the variance of pollutants as a function of time or meteorological factors.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
Errata sheet for EPA report EPA-600/4-78-027
"Study of the Subarctic Heat Island at Fairbanks, Alaska"
p. 11 line 29: reference should be to Table 3 section 5_.
p. 67 caption Figure 36: Z = 45 ra
p. 94 last line should be moved up to become 113 line 7; also last
word of that line should be radius.
443
p.113 polynomial expansion in line 13 fi2 should start oTrg = alT + 4TQ
p.125 eqn 23: F should be <
p.127 footnote: T = .046
p.132 eq 36: right hand side should be y
. / 2^
p.133 eq 46: should be * = 1 + yU V^ + [y]
I '
-1
-7
p. 140 Table 24 2nd column: heading should be "YJ/K = |,
-------
EPA-600/4-7R-027
June 1978
STUDY OF THE SUBARCTIC HEAT ISLAND
AT FAIRBANKS, ALASKA
S. A. Bowling and C. S. Benson
Geophysical Institute
Fairbanks, Alaska 99701
Grant No. 80299
Project Officer
George C. Holzworth
Meteorology and Assessment Division
Environmental Sciences Research Laboratory
Research Triangle Park, North Carolina 27711
ENVIRONMENTAL SCIENCES RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 27711
-------
DISCLAIMER
This report has been reviewed by the Office of Research and Monitoring,
U. S. Environmental Protection Agency, and approved for publication.
Approval does not signify that the contents necessarily reflect the views
and policies of the U. S. Environmental Protection Agency, nor does mention
of trade names or commercial products constitute endorsement or recommendation
for use.
-------
ABSTRACT
The heat island associated with the city of Fairbanks, Alaska was
studied as a means of isolating the effects of self-heating and modified
radiative transfer from other causes of heat islands. Minimal winter
insolation virtually eliminated the effects of variable albedo and the
daily temperature cycle; snow cover and dormant vegetation made differences
in evapotranspiration unimportant, and very low wind speeds minimized
the effect of surface roughness.
The observed steady-state heat island under clear skies and low wind
speeds was around 10°C, with transient values reaching 14°C. This high
value is probably due to the extremely steep ground inversions known to
exist in Fairbanks, as the heat island intensity correlated well with
the strength of the inversion between 2 and 60 meters elevation. The
depth of the mixing layer was less than 90 meters, but the temperature
structure at higher levels was disturbed, apparently by coherent lifting
of the stable air. The mean surface wind field was extremely complex in
both time and space, with strong vertical shears, horizontal eddies with
scales from a few hundred meters to several kilometers, and seiche
oscillations at several scales superimposed on gravity drainage. Speeds
were generally too low for accurate measurement.
A self-heating term of 10KW person' in winter and 5KW person" in
summer was derived from the fuel inventory carried out as part of the
project. The winter value, applied in a simple model of a heat island
over a conducting and radiating city, gave realistic heat island values
with wind speeds under 1 m sec~ .
iii
-------
CONTENTS
Abstract Hi
Figures v
Tables xv1
Acknowledgement xv11i
1. Introduction 1
2. Conclusions 4
3. Recommendations 6
4. Background Information 8
The geographic setting 8
Historical data 11
5. The Surface Temperature Field 17
Measurements 17
Results 35
6. The Three-Dimensional Heat Island 61
Measurements 61
Results 63
7. The Wind Field 72
Measurements 72
Results 75
8. Fairbanks Energy-Use Inventory 85
Electricity 85
Coal 85
Gasoline 89
Fuel oil 94
Total energy 100
Breakdown of energy use 101
Production of unavoidable combustion products 108
9. Theory 110
Introduction 110
The effect of the form of the temperature profile- ... 119
Conductive/convective and radiative energy losses. ... 131
References 147
-------
FIGURES
Number Page
1 Fairbanks and surroundings as seen by LANDSAT. The outlined
area is that shown in Figure 2. North is at the top 9
2 Map of the Fairbanks area, showing thermograph sites and traverse
routes. Routes most often used were University Avenue, Airport
Road, Cushman Street and College Road. Letter codes are as follows:
Thermographs: AL Alaskaland, CF Creamer's Field, CL City Parking
lot, FS Fire Station, FW Farm Woods, LF Lower Farm, PR Peger Road,
SR Sears Roebuck Co, UF Upper Farm (=University Exp. Station
climatological station). Creamer's Field and Peger Road had wind
data available during part of the second year. Other data: WS
NOAA Weather Service, A site of acoustic sounder/boundary layer
profiles study, B Borough-operated wind sensor, N anemometer at
North Slope Batteries. Thermograph sites CL and SR were "downtown",
FS and AL were suburban, and the remaining sites were rural.
Shading indicates elevation, with unshaded areas 130 to 152m MSL,
lightly shaded areas 152 to 305m MSL and heavily shaded areas
above 305m. North is at the top of the figure 10
3 Fairbanks heat island, 13 December 1964 (Traverse V). Contours labeled
in °C 12
4 Differences in monthly mean temperatures for November, downtown
Fairbanks minus the Experimental Farm. All years but 1933 had
downtown thermometers at 3m height; 1933 was 20m temperatures.
Negative temperature differences in most years are due to intense
inversions (Experimental Farm was 18m higher than downtown) 16
vi
-------
FIGURES
Number Page
5 Summary of weather and data collection activity, Nov. 1974
through April 1975. From top to bottom: Double line graph
gives daily maximum and minimum temperatures. Dots in line
with TR show traverse times; x's in line with FL indicate
aircraft observations. Under wind, W's indicate multiple-
observer wind observations; single lines indicate one wind
recorder in operation, double lines indicate two or more
recording anemometer records available. Sky conditions:
filled curve gives sky coverage by cloud in tenths; open
curve gives sky obscuration by ice fog. Horizontal lines:
periods of thermograph records; code letters are the same as
in Figure 2 18
6 Same as 5 but for May, 1975 through October, 1975 19
7 Same as 5 but for November, 1975 through May, 1976 20
8 Locations of thermistor sensors on traverse vehicle, (a) side
view; (b) front view; (c) detail of thermistor shield (used for
late-spring and summer daytime runs.) 24
9 Vertical view of downtown Fairbanks, showing the heat island core
area (white arrow) and the thermograph site CL, 21 April 1975 ... 27
10 Thermograph site CL, looking south-southeast, in late April, 1976.. 28
11 Vertical view of Creamer's Field, College Road at lower edge,
showing thermograph site CF, 21 April 1975 29
12 Thermograph site CF, looking NNE, April, 1975 30
vii
-------
FIGURES
Number Page
13 Vertical view of Peger Road area, 21 April 1975, Tanana River at
lower part of photo. Peger Road thermograph locations marked by
PR 31
14 Original site PR looking west from access road, summer 1975 32
15 Both PR sites, photo taken in April 1976 from the same location as
was Figure 14. The tallest spruces at the left mark the old site;
the thermograph shelter is at the new site 33
16 Both PR sites, looking SE, anemometer at the new site in the
foreground 33
17 Thermograph temperatures, night of March 3-4, 1975 (23). Stations
keyed to Figure 2. Note that station UF is about 12 m higher than
the other stations 34
18 Temperature contours at 2m elevation for a typical heavy ice-fog
situation, 1300-1400 January 3, 1975 (8). Contours labeled in °C. . 36
19 Aerial view of Fairbanks during traverse 8, 3 January 1975 37
20 Temperature contours at 2m elevation for Traverse 13, 1600-1900
January 10, 1975. Moderate ice-fog 38
21 Temperature contours at 2m elevation for Traverse 29, 2000-2200
December 1, 1975. Moderate ice-fog 39
22 Comparison of low-level sounding data for times near Traverses 8,
(1/3/75), 11 (1/7/75), 13 (1/10/75), all based on rawinsonde data,
and traverse 29 (12/1/75), based on helicopter data. Traverse 11
had the weakest nighttime heat island in the study 40
viii
-------
FIGURES
Number Page
23 Temperature contours at 2m elevation for Traverse 24, 2320 13-0120 14
March, 1975. Clear skies 42
24 Temperature contours at 2m elevation for Traverse 38, 2100-2230
February 26, 1976. Clear skies. Measured wind directions shown by
heavy arrows; wind directions inferred from plume drift shown
by light solid arrows (ground level) or light dashed arrows
(high stack plumes) 43
25 Temperature contours at 2m elevation for Traverse 17, 1430-1600
January 25, 1975. Overcast skies. Arrows show approximate mean
wind directions over the half hour period 1440-1510 44
26 Thermal infrared imagery of the Alaskaland (AL) area, 0500
March 4, 1975. The general gradient of density from light at
the bottom to dark at the top is an artifact of processing and
should be ignored. (Photo courtesy of U. S. Army 172nd MID(AS)
and Cold Regions Research and Engineering Laboratory) 49
27 Thermal infrared image of part of downtown Fairbanks, about 5 a.m.
on a clear March morning. Note the importance of the river as
a heat source, especially downstream of the Municipal Utilities
System power plant near the center of the picture. The general
gradient of density from light at the bottom to dark at the top
is an artifact of processing and should be ignored. (Photo
courtesy of U.S. Army 172nd MID(AS) and Cold Regions Research and
Engineering Laboratory) 49
28 An example of daily temperature variations in town (solid line,
thermograph CL) and at Creamer's Field (dashed line) very early
in the snowmelt season. Skies were clear throughout 51
ix
-------
FIGURES
Number Page
29 Comparison of daily minimum temperatures at four locations in March.
Heavy solid line is site CL (downtown), light solid line is the
official weather service temperature recorded at the airport (WS),
dashed line is Creamer's Field (CF) and dotted line is Peger Road
(PR), both outlying stations. Squares show sky conditions, with
open squares for clear skies, black squares for complete overcast
and intermediate shading for intermediate conditions. Slant of
shading indicates increasing (up to right) decreasing (down to
right) or constant (crossed lines) cloudiness 52
30 Oblique aerial photograph of Fairbanks, Alaska, looking north,
taken during the snowmelt season, April 21, 1975. Note the lack
of snow on roads and in the city core area just south of the wider
bridge (Cushman Street). For comparison with Figure 2, the broad
E-W highway near the bottom of the figure is Airport Road and the
road bordering the snow-covered field at top is College Road. The
Creamer's Field thermograph is at the end of the side road north
near the west edge of the subdivision roads south of College Road.. . 54
31 An example of daily temperature variations in town (solid line)
and at Creamer's Field (dashed line) quite late in the snowmelt
season. The curve at the bottom of the figure gives cloud cover
in tenths 55
32 Temperatures and cloud cover during a relatively clear period in
early July, 1975. Stations are the same as Figure 29 56
33 Thermal infrared imagery of the Fairbanks area in summer. Site
CF is at the top edge. This scene was dated to approximately
summer 1970 largely on the state of completion of the intersection
of Airport Road and Cushman streets, near the lower edge 57
-------
FIGURES
Number Page
34 Vertical profiles of temperature, acoustic sounder backscatter
and wind velocity in a relatively undisturbed area (From Holmgren
et. al., 1975) 63
35 Comparison of heat island intensity and approximate 60 m inversion
strength (Geophysical Institute temperature minus the same back-
ground temperature used in computing the heat island intensity). . . 65
36 Plot of crossover height against inversion strength based on
empirical relationship between heat island intensity and 60m
inverson and assumed constant background lapse rate.
Assymptotic approach to Z = 45m for strong inversions is
physically unrealistic and indicates probable non-constant back-
ground lapse rates for these conditions 67
37 Limiting values for city lapse rate. Heavy dots are measured
temperatures; light line is the adiabatic lapse rate over the city;
dashed line is the background temperature profile assuming a constant
lapse rate to 60m; dash-dot line is a more realistic background
temperature profile. Cross shows calculated crosssover height,
circled cross, a more realistic value. The true city temperature
profile must lie within the triangle formed by the heavy solid lines. 68
38 Thermistor traces across town during Traverse 23 at an elevation of
90m. Left side - North to South and return along Cushman Street;
right side - East to west along the line of Airport Road 69
39 North-South temperature cross section of the air over Fairbanks,
based on data obtained midnight to 0100 March 14, 1975 (Traverse
24) 70
XI
-------
FIGURES
Number Page
40 Photo of the Fairbanks area, looking east, taken 3 January 1975
with a temperature of -42.5° measured at the city core. East,
west, and north winds may be observed to affect plumes emitted
at different levels 74
41 Comparison of winds at three sites through the course of a day. ... 76
42 Suites of wind directions for various simple wind fields for
comparison with Figure 41. Wind measuring sites coded same as
Figure 41 76
43 Wind vectors for 2 two-minute measuring periods during Traverse 20
(0900-1000 8 February 1975). The lengths of solid arrows indicate
measured wind speeds; dashed arrows indicate that only wind
direction was available. Filled arrowheads-0925; open arrowheads
1000 77
44 Low-level wind roses at sites in the Fairbanks area for 4 periods
of light ice fog in 1969-1970 78
45 Surface flow pattern for ice fog deduced from the data in Figure 44. 82
46 Cumulative commerical electricity production as a function of time
in and for the Fairbanks area. Note that the top line is the sum of
all commercial generation in the area 87
47 Cumulative coal consumption by power plants in the Fairbanks area
in thousands of tons 88
48 Gasoline consumption in the Fairbanks area. Heavy line - adjusted tax
data; light line - Alaska Railroad imports 92
xi i
-------
FIGURES
Number Page
49 Gasoline imported to Fairbanks and Fort Walnwight prior to 1965
(not cumulative) 93
50 Cumulative fuel oil use in the Fairbanks area 100
51 Fuel oil imported to the Fairbanks area prior to 1965 (not
cumulative) 101
52 Total cumulative energy use in the Fairbanks area broken down by fuel
type and compared with monthly mean temperatures 105
53 Surface and fog-top energy balances for (a) clear background
area; (b) clear city; (c) foggy city 114
54 Cross section through a hypothetical city, wind from left to right
in all cases. Panel a: assumed heat input, h, and heat transfer-
red to the air, q. Panel b: Values of AT and observed AT
for the given values of h and q. Panel C: Height of mixed layer,
Z, and development of inversion downwind of maximum heat island
under clear conditions. Panel d: Height of mixed layer, Z, and
cooling downwind of maximum heat island under foggy conditions. . . .117
55 City vs. background soundings for clear and city-fog cases. Both
city soundings have had the same net heat addition, q, which is in
each case proportional to the shaded area between the unmodified
and modified soundings. The fog case has some negative area
(horizontally shaded) due to cooling from the fog top, which is
balanced by an increase in the positive area, the increase being
shown by the fine vertical lines.
A6p = ATp = Heat Island intensity for the clear case
A0r = ATp = Heat Island intensity for the foggy case 119
XTM
-------
FIGURES
Number Page
56 Development of mixing layer as background air moves into the city.
y1 is the particular value of y selected for the right hand panel.
Left panel - background sounding. Center panel - cross section
showing increase of Z (for non-uniform q) as air moves into city.
Right panel - sounding at y = y1 within the city. Shaded area -
Z(o(z) - o(z))dz is equal to Q(y) 120
0
57 Potential temperature profiles for which calculations were made of
mixing height and heat island intensity. Light lines - lapse rates
with the 100 meter potential temperature difference equal to
15.22°C. Plain line - linear case; line with crosses - logarithmic
case; dashed line - stepped case; dotted line - capping inversion
case. Heavy lines - linear and logarithmic profiles for 100 meter
potential temperature differences of 4.6°C (3.6°C/100 m inversion). 128
58 Mixing height, Z, as a function of accumulated bottom heating,
Q(y), in °K m. Line types as in Figure 57. Short double lines
indicate the 100 m level 129
59 Heat island intensity, AS, as a function of Q(y). Line types as
in Figure 57. Short double lines indicate where the heat island
extends above 100 m 130
60 Development of city fog sounding 139
61 Dependence of city center heat island intensity on population. Heavy
lines include the effect of thermal radiative loss, light lines
give heat island intensity neglecting radiative effects 144
62 Dependence of city center heat island intensity on energy release
per person (solid lines) and potential lapse rate (dashed lines).
Light and heavy lines as in Figure 61 144
xiv
-------
FIGURES
Number Page
63 Dependence of city center heat island intensity on wind speed
(solid lines) and area per person (dashed lines). Light and heavy
lines as in Figure 61 145
xv
-------
TABLES
Number Page
1 Monthly mean differences in winter daily minimum temperatures
between Experimental Farm and Weeks Field Stations before and
after Experimental Farm move 14
2 Comparison of the Fairbanks winter heat island in 1929-42 with
that in 1974-76 15
3 Summary of Traverse Data 21
4 Day length and maximum solar elevation at Fairbanks as a function
of time of year .46
5 Thermal properties of various substrates found in the Fairbanks area.48
6 Approximate height of Fairbanks heat island as a function of
inversion strength 66
7 Electricity imported to Fairbanks 86
8 Fairbanks area coal usage by year 90
9 Monthly power plant coal use breakdown in 10 tons - 1974 and 1975. .91
10 Total coal use in the Fairbanks area 92
11 Fairbanks area gasoline consumption 95
12 Ratio of taxed yearly sales of diesel oil to those of gasoline
for the IV Judicial District 96
xvi
-------
TABLES
Number Page
13 Fairbanks area fuel oil consumption 98
14 Total energy use in the Fairbanks area 102
15 Average power consumption per person (including waste heat) in
the Fairbanks area for 3 winter and 2 summer months 106
16 Breakdown by point of release of energy used in the Fairbanks area. .107
17 Production rates of H20 and C02 by fuel types 108
18 H20 and COo generated by fossil fuel combustion in the Fairbanks
area in 1975 109
19 Symbols used J21
20 Temperatures from two helicopter traverses J30
21 K/J as a function of AB/J J33
22 Typical values of A and B 334
23 Predicted heat islands ]37
24 Maximum ratios of foggy/clear values of mixing depth (/ J/K) and
AT (m) J40
25 Combination of 4., G, and K observed for urban areas ]43
26 Variation of AT with population, area per person, energy
release per person, lapse rate and wind speed ]46
xvii
-------
ACKNOWLEDGMENTS
Section 8 of this report is based on data provided by a number of
organizations and businesses. These included:
Alaska Dept. of Revenue
Alaska Railroad
Federal Energy Administration
Fort Wainwright Power Plants
(U.S. Army)
Golden Valley Electrical Assn.
Municipal Utilities System
Univ. of Alaska Heating Plant
Alyeska Pipeline Company
Fairbanks Fuel Supply
Johnny's Express
Kobuk Oil Company
Northern Heating Oil
Peters Express
Petroleum Sales Limited
Shell Heating Oil
Sourdough Heating Inc.
We would like to thank all of the above, and to express our special
gratitude to the power and heating plants, which provided pre-1967 data,
and to the managers of Sourdough Heating and Fairbanks Fuel Supply, who
supplied the long-term monthly data which allowed us to extend the data
from other oil suppliers backward in time.
xviii
-------
SECTION 1
INTRODUCTION
"Urban heat island" is a term used to indicate the effect of a city on
the temperature measured within that city. Its existence has been noticed
since 1833, and various detailed studies have been made over most of the
last 50 years (Duckworth and Sandberg, 1954). Ideally, the magnitude of
the heat island is the difference between the temperature measured within a
city and the temperature which would be measured at the same point if the
city were not present. In practice, the temperature difference between the
city and its surroundings is used. Heat islands may be positive or negative,
although the negative type is rarely considered.
The most obvious effect of a heat island is, of course, the direct
effect on temperature. Hot summer nights are even hotter in large cities.
But probably the most compelling immediate reason for studying the urban
heat island is the interaction between the heat island and air pollution.
Pollutant concentrations are critically dependent on local air motion and
temperature structure. In order to develop realistic models for pollutant
dispersal, the heat island structure and its dependence on conditions that
may change with time must be considered.
A heat island is produced by a combination of many factors, which
often oppose each other (Myrup, 1969). The major factors leading to an
increase in city temperature are: (1) the heat exhausted to the atmosphere
from energy sources within the city (anthropogenic heat or self-heating),
which might be considered analogous to thermal pollution of surface waters,
(2) a decrease in the area of transpiring plants within the city, (3) The
high heat capacity and conductivity of building materials and the generally
increased roughness of cities over their surroundings tend to decrease the
amplitude of the daily temperature wave (Nappo, 1972). (This will tend to
1
-------
increase further the night-time temperature, but decrease the temperature at
midday.) (4) The radiative effects of dirty city air also have an influence
on the city air temperature (Atwater, 1971), and if the city albedo differs
from that of its surroundings, this will also have an effect.
A high-latitude city, such as Fairbanks, Alaska, provides separation
of these causative factors, especially in winter. Once the winter snow
cover is established (usually sometime in October) differences in evapora-
tion become unimportant. Even where snow is removed from the main roads
enough of a veneer of ice remains that saturation of air in contact with
the surface is assured. During December and the first half of January the
sun is above the horizon for less than five hours a day, with a maximum
elevation above the horizon of only about three and a half degrees. (The
minimum values are 3 1/2 hours daylight with noon-time solar angle of
1.5°.) With so little incoming solar energy and the presence of a high-
albedo snow surface to reflect most of this slight amount back to space, it
is not surprising that the "normal" daily rise in temperature cannot be
detected without statistical techniques. This in turn means that the
difference in heat capacity and conductivity between the city and the
surrounding contryside affects the heat island intensity only during
changes from one type of weather to another. The very low wind speeds
typical of Fairbanks in winter also minimize the effect of differing
roughness in and out of the city. On the other hand, Fairbanks is subject
to high concentrations of several pollutants, with one component of the air
pollution -- ice fog -- having particulary marked effects on radiative
transfer. In addition, the per capita energy consumption in winter is
comparable to that of the heavily urbanized strip along the east coast of
the U. S. Thus the Fairbanks heat island in winter is dependent primarily
on anthropogenic heating, with ice fog being present at some times.
As the seasons progress, other factors begin to influence the Fairbanks
heat island. By February, a normal daily temperature cycle is present but
melting is still generally absent. Substrate properties become important,
but evapotranspiration can still be neglected. The latent heat of melting
snow - a factor not normally considered in heat island studies - becomes
important in April. By mid-May to June vegetation is in full and extremely
-------
rapid growth under uninterrupted daylight and approximately twenty hours a
day of possible sunlight. At the same time, the anthropogenic heating has
decreased to about half of its winter value. Thus in summer, conditions are
similar to those normally encountered in temperate latitudes.
The following report details an observational and theoretical study of
the Fairbanks heat island. This information should prove useful in evaluating
the relative importance of the causative factors in other areas as well.
-------
SECTION 2
CONCLUSIONS
Our investigations have demonstrated that the Fairbanks area, with an
estimated population of 65,000, frequently has a winter heat island intensity
of 10°C, and intensities of up to 14°C have been observed. Thus self-
heating with some disturbance by buildings of thermal infrared transfer is
adequate to produce a substantial heat island. The magnitude of the night-
time heat island is only slightly less in the summer months. The daytime
heat island disappears from March to September except for a short period in
late spring when snowmelt is complete in the city but not in the surrounding
area. The effect of ice fog is to decrease the heat island intensity.
This effect is probably due to the fact that ice fog affects the background
areas as well as the city.
The Fairbanks heat island appears to extend to roughly 60m above the
city. Temperatures at 90m are warmer than those at street level under
clear and slightly foggy conditions, but greater mixing depths may exist in
dense fog. The heat island intensity correlates well with the intensity of
the inversion in the lowest 60m of the atmosphere.
The wind field was found to be extremely complex, with 180° shears
being common over distances of 200m in the horizontal and 25 m in the
vertical. Fluctuations with time are also substantial. It was not possible
to determine unequivocally the effect of the heat island on the wind field.
Fairbanks is hardly an industrialized area, but the fuel use inventory
gave per capita winter energy consumptions of the order of 10 KW /person.
This is comparable to Los Angeles or the east coast urbanized strip from
Boston to Washington, D. C. (SMIC, 1971). Summer energy use is about 5
KW/person.
-------
A simple theoretical analysis, including both radiative and convective
losses from the surface, gives fairly good agreement with the observed
winter heat island, but only if regional wind speeds are kept well below 1
m/sec. This figure is compatible with our observations of the wind field.
In summary, our investigations have demonstrated that a strong heat
island can exist even when the only contributing factors are self heating
and some disturbance of radiative transfer due to the presence of sizeable
buildings. The heat island thus produced is not, however, strong enough
to break the Fairbanks inversion even up to 100 m, at least under clear-
night conditions.
5
-------
SECTION 3
RECOMMENDATIONS
Although the horizontal distribution of temperature around and in
Fairbanks and its variation with season and time of day are now reasonably
well understood, some questions remain unanswered. These problems concern
the wind field and the detailed vertical temperature structure within the
downtown area, both of which are of major inportance for any continued air
pollution studies in the Fairbanks area. The vertical temperature structure
outside the city has been examined in detail in the past (Holmgren et al,
1975).
As pointed out in Section 7, the wind field is now known to be extremely
complex, and a complete formulation based on any number of point measurements
at any one time would be of questionable utility even if it could be achieved.
On the other hand, the construction of a refinery at North Pole (about 20 km
ESE of Fairbanks) and the proposed further industrialization of that area
have made an understanding of the wind fields actually involved in pollutant
transport of immediate practical importance. We strongly recommend that
tracer studies be carried out in the area. If the emissions from the North
Pole refinery prove to have a sufficiently unique trace element "fingerprint",
they may themselves act as an automatically released tracer. Their usefulness
should be confirmed by an inert tracer such as SFg or a Freon not normally
used in the area, which could also be used to check flow from possible
source areas other than North Pole. The manager of the North Pole refinery
has indicated a willingness to cooperate in such studies. As an adjunct to
such a study, the time-lapse photography of vapor plumes carried out experi-
mentally during the heat island project and more intensively during the
1976-1977 winter should be continued and expanded.
-------
Another portion of the study which was not as complete as we would have
liked was the vertical temperature profile within the city. It appears
likely from the helicopter measurements that the vertical temperature
structure within the city, studied in detail, might show horizontal variabil-
ity approaching that of the wind field. (In fact the vertical structure must
be tied to the convergence of the wind field by simple conservation of mass).
Pilot ballons and a thermistor are not adequate (see Section 6). Radio-
controlled model airplanes, also considered, will not operate reliably at
temperatures below -20°C. Probably the best way of studying the 2-100 m
region would be something like the Institute's boundary layer profiling
(BLP) system (part or all of which was unfortunately in Antarctica during
our study) but mounted in a trailer or truck. This system consists of an
3
instrument package lifted to as much as 500 m by a 3.2 m aerodynamic balloon,
with temperature, pressure, and wind speed telemetered back to the ground.
It has previously been used in a stationary mode in the flat, undeveloped
area between Fairbanks and the University of Alaska. Some results from the
system are cited in Section 6. We would recommend the use of such a system
in conjunction with mobile measurements of air chemistry currently in the
planning stage.
-------
SECTION 4
BACKGROUND INFORMATION
THE GEOGRAPHIC SETTING
Fairbanks is located on the floodplain of the Chena River near its
confluence with the Tanana River. South of the town site, very gently sloping
outwash plains extend roughly 100 km from the Tanana River to the Alaska
Range. From southwest through northwest to northeast of Fairbanks proper,
residential areas extend up into the Tanana-Yukon uplands. East and southeast
of town the uplands are farther away (30 to 50 km), but even at this distance
the influence of the hills on the wind field cannot be neglected under strong
inversions. Figure 1 shows the topography of the area as seen from space,
while the smaller scale topography and the locations of thermograph sites and
roads used for traverses are shown in Figure 2.
Because of the intense winter ground inversions, (Benson, 1970), topo-
graphy has an overwhelming influence on both the wind and the temperature
fields in the area, as will be discussed in Sections 6 and 7. In considering
background temperature sites comparable to the downtown area (which is right
on the river bank) any area which is in the hills may be eliminated at once.
Luckily the break in slope between the flood plain (locally referred to as
"the flats") and the hills is usually abrupt. Even within the flats, however,
filled-in sloughs and similar very slight depressions may be considerably
colder than nearby areas which are not obviously higher. The infrared images
reproduced in Section 5 make this quite evident.
In considering how the site on which Fairbanks is built would fit into
the natural temperature field if the city were not there, two types of infor-
mation were considered. The first was the distribution of water depths
-------
Figure 1. Fairbanks and surroundings as seen by LANDSAT. The outlined area
is that shown in Figure 2. North is at the top.
during the Fairbanks Flood of August 1967 (Childers and Meckel, 1967).
Virtually everything within the city limits of Fairbanks and a large area
outside the city were under 2 to 5 feet of water. Of the background thermo-
graph sites selected, Creamers Field was under approximately 2 feet of water,
Peger Road under 3 feet, and Lower Farm was out of the flooded area. The heat
island core at Second Avenue and Cushman Street was under 2.7 feet of water.
Thus to the extent that water flows in the same way as cold air, the back-
ground sites should have temperatures similar to or higher than that of the
heat island core. (It is probably worth noting that the Airport, where
weather records are now taken, was an island above the flood waters.)
-------
\ jT~\ ' 7 ^w_
Figure 2. Map of the Fairbanks area, showing thermograph sites and traverse
routes. Routes most often used were University Avenue, Airport
Road, Cushman Street and College Road. Letter codes are as follows:
Thermographs: AL Alaskaland, CF Creamer's Field, CL City Parking
lot, FS Fire Station, FW Farm Woods, LF Lower Farm, PR Peger Road,
SR Sears Roebuck Co, UF Upper Farm (=University Exp. Station
climatological station). Creamer's Field and Peger Road had wind
data available during part of the second year. Other data: WS
NOAA Weather Service, A site of acoustic sounder/boundary layer
profiler study, B Borough-operated wind sensor, N anemometer at
North Slope Batteries. Thermograph sites CL and SR were "down-
town", FS and AL were suburban, and the remaining sites were
rural. Shading indicates elevation, with unshaded areas 130 to
152m MSL, lightly shaded areas 152 to 305m MSL and heavily shaded
areas above 305m. North is at the top of the figure.
In addition, the lowest temperature ever recorded by the weather service
in the Fairbanks area, -66°F (-54.4°C) on January 14, 1934, was recorded at
Second and Cushman--the heart of the present day heat island. As discussed in
the next subsection, this temperature was itself probably influenced by the
10
-------
Fairbanks heat island. A quick check of other interior Alaska stations in
operation since 1934 indicated that the 1934 cold spell, although certainly
severe, has been equalled or exceeded since—but not in Fairbanks, where the
weather station has been moved to other sites. We concluded that the natural
background temperature at Second and Cushman was probably comparable to those
observed at our colder background sites.
The final choice of Creamers Field rather than Peger Road as the primary
background site was based on winds. Although the wind field in the Fairbanks
basin is extremely complex, topography and strong radiative cooling do drive
a very slow downslope motion (north to south) at low levels. Seiche oscil-
lations, the influence of the regional wind field, Coriolis effects on the
downslope winds, and local eddies and waves would exist even in the absence of
the city (see Section 7), but only the gravity drainage (modified by Coriolis
effects) produces a consistent direction of transport. As a consequence,
Creamers Field, which is north of town, is usually sampling the clean air
coming into the town while Peger Road air has often been through Fairbanks.
The distribution of ice fog, which under light to moderate fog conditions was
normally absent at Creamers' Field and present at Peger Road, confirms this.
(Creamer's Field is actually closer to heavy traffic.) Under heavy fog
conditions Creamer's Field is often fogged in also, and in fact the apparent
weakening of the heat island under these conditions (see Section 5) may be
because we have been unable to locate an accessible background site in the
flats upwind of and uninfluenced by the City of Fairbanks.
HISTORICAL DATA
Most heat island studies are confined to data collected after the study
is started. In Fairbanks, data from two earlier periods are also available.
Benson (1970) made several temperature traverses under ice fog and partially
overcast conditions during the course of his ice fog investigations in the
early 1960's. (See Roman-numbered traverses in Table 3, Section 4.) In
addition, some evidence for heat island intensity is available from the
1930's.
11
-------
Benson's traverses generally showed temperatures similar to those observed
during our ice fog measurements in 1974-1976, except that the secondary
maximum temperature at Cushman and Airport Road was not so pronounced during
the earlier study. Figure 3 is a plot of one of these early traverses, which
may be compared with plots at similar temperatures (Figures 18, 20, 21, 23,
and 24) in Section 5. His traverses at higher temperatures showed generally
weak heat islands.
The data for the 1930's comes from the fact that Fairbanks has had both
a Weather Bureau Station and a climatological station at the University
Experimental Farm since the summer of 1929. Although both stations have been
moved, there is enough common data that we can compensate for the experimental
Fiaure 3. Fairbanks heat island, 13 December 1964 (Traverse V).
labeled in °C.
12
Contours
-------
farm move in the late 1940's. It is then possible to compare the temperature
differences between the two stations in the 1930's (when the Weather Bureau
was located in the heart of downtown Fairbanks, within a block of the present
heat island maximum at Second and Cushman) with those available for the
period when thermograph records were available from downtown during 1974-76.
Compensation for the change in Experimental Farm location is based on the
period from 1943 to 1951 when the Fairbanks station was located at Weeks Field
(north of Airport Road and west of Cushman Street; see Figure 2). Prior to
1943 the Fairbanks station was at Ladd Air Force Base (now Ft. Wainwright) or
downtown; since 1951 it has been at the present International Airport. The
Climatological Record for Alaska shows that the University Experiment Station
changed from 500' (152 m) to 475' (145 m) msl elevation in the summer of 1947.
Internal evidence in the temperature records suggests that the actual movement
of the station may have taken place in the summer of 1946, so the 1946-47
winter was excluded from the analysis. Both the old and the new station
locations are on an exposed south slope, the new location being the one marked
UF in Figure 2. (By comparison, the City of Fairbanks -is at 440' [134 m]).
In order to avoid insolation effects, only December and January monthly means
of daily minima were used. Temperature differences between the two stations
for the three years before the move and the four years after are given in
Table 1. If the Weeks Field location remained unchanged, comparison of means
for the two periods gives a 2.5°C temperature change due to the move of the
Experimental Farm. If the apparent trends within the two short periods are
real and due to an increased effect of the Fairbanks heat island on the Weeks
Field location, the change could be as little as 1.5°C or even 1°C. We have
used the 2.5°C correction in Table 2, recognizing that the corrected values
could be as much as 1.5°C too high. The Experimental Farm location itself has
been changed by paving the road 100 m to the south and by addition of a small
modular building south of the measurement area. Possible effects of these
changes have been neglected.
As the Experimental Farm site is definitely higher and warmer than the
flat area around Fairbanks, a final correction was made by using the available
data from Creamer's Field for the 1974-76 December and January minimum temper-
atures. The resulting mean heat island intensities (with the 1° correction
for the location of the downtown thermograph) are given in Table 2 for the
13
-------
Table 1. MONTHLY MEAN DIFFERENCES IN WINTER DAILY MINIMUM
TEMPERATURES BETWEEN EXPERIMENTAL FARM AND WEEKS
FIELD STATIONS BEFORE AND AFTER EXPERIMENTAL FARM
STATION MOVE. POSITIVE VALUES INDICATE THAT THE
EXPERIMENTAL FARM WAS WARMER THAN WEEKS FIELD.
500' location
475' location
Dec 43
Jan 44
Dec 44
Jan 45
Dec 45
Jan 46
Dec mean
Jan mean
Dec-Jan
Change:
3.9
3.3
2.4
3.7
3.0
2.1
3.1
3.0
means 3.1
~9 CjOf
ut. . _; o
Dec 47 0.7
Jan 48 1.9
Dec 48 0.3
Jan 49 0.7
Dec 49 0.1
Jan 50 -0.3
Dec 50 0
Jan 51 -0.2
Dec mean
Jan mean
Dec-Jan means
0.3
0.5
0.4
recent data. The difference between the temperatures from the Experimental
Farm and those from Creamer's Field ranged from 2 to 4.5°C, averaging 3.3°C.
In order to estimate the heat islands of earlier years, 3.3°C was added to the
site-corrected figures for 1929-42. The results are shown in Table 2.
Census populations were 2100 for 1930 and 3450 for 1940, as compared with
18,400 for 1970; so the change from the thirties to the mid-seventies is in
fact very nearly proportional to the change in the fourth root of population
(see Section 9).
One other point of interest may be gleaned from the records of the 1930's,
In 1933, when the Weather Service moved from First and Cushman to Second and
Cushman, the thermometer, previously maintained at a height of 3 m, was in-
stalled at a height of 20 m from July through November. By late November, it
had become obvious that the change in height was affecting recorded tempera-
tures and the thermometer was moved back to a 3 m height. Figure 4 shows the
14
-------
TABLE 2. COMPARISON OF THE FAIRBANKS WINTER HEAT ISLAND IN 1929-1942 WITH THAT
IN 1974-1976. ALL FIGURES ARE DIFFERENCES (IN °C) BETWEEN MONTHLY
MEANS OF DAILY MINIMUM TEMPERATURES, CITY MINUS EXPERIMENTAL FARM.
1929-1933 CITY TEMPERATURES FROM FIRST AND CUSHMAN, 1933-42 FROM
SECOND AND CUSHMAN. 1974-1976 FROM BARNETTE (TWO BLOCKS WEST OF
CUSHMAN) BETWEEN THIRD AND FOURTH. SECOND AND CUSHMAN IS CURRENTLY
THE PRIMARY MAXIMUM OF THE HEAT ISLAND, WITH THE BARNETTE STREET
LOCATION BEING 2°C COLDER DURING STRONG HEAT ISLAND CONDITIONS AND
AN ESTIMATED 1°C COLDER ON A MONTHLY MEAN BASIS. STATION LOCATION
CORRECTIONS ARE +2.5°C PRIOR TO 1947 (CORRECTION FOR EXPERIMENTAL
FARM SITE CHANGE) AND + 1° IN 1974 IN 1974-1976 (CORRECTION FOR
THERMOGRAPH SITE LOCATION RELATIVE TO SECOND AND CUSHMAN).
December
Year
1929-30
1930-31
1931-32
1932-33
1933-34
1934-35
1935-36
1936-37
1937-38
1938-39
1939-40
1940-41
1941-42
Mean
1974-75
1975-76
Mean
Raw Differences
-1.9
-2.1
-1.4
-2.5
-0.8
-2.5
-0.8
-0.9
-2.2
-0.8
0.9
0.9
-2.6
2.8
2.6
Corrected for
Site Changes
0.6
0.4
1.1
0
1.7
0
1.7
1.6
0.3
1.7
3.4
3.4
-0.1
nr?
3.8
3.6
~3T7"
Corrected for
Background
Temperatures
3.9
3.7
4.4
3.3
5.0
3.3
5.0
4.9
3.6
5.0
6.7
6.7
3.2
4T5"
8.2*
5.9*
7HJ
1929-30
1930-31
1931-32
1932-33
1933-34
1934-35
1935-36
1936-37
1937-38
1938-39
1939-40
1940-41
1941-42
Mean
1974-75
1975-76
Mean
-3.6
-2.7
-2.3
-2.3
-1.3
-7.9
-3.2
0.7
-2.1
-0.9
-1.5
-0.3
-3.1
1.0
2.7
-1.1
-0.2
0.2
0.2
1.2
-5.4
-0.7
3.2
0.4
1.6
1.0
2.2
-0.6
~0
2.0
3.7
~FS
2.2
3.1
3.5
3.5
4.5
-2.1
2.6
6.5
3.7
4.9
4.3
5.5
2.7
TT
6.6*
5.6*
6TT
* Measured values (downtown-Creamer's Field thermograph, +1° correction
for downtown thermograph site). These are the values used to estimate
the 3.3°C correction (2° to 4.5°) to the earlier data.
-------
-3-
YEAR
Figure 4. Differences in monthly mean temperatures for November, downtown
Fairbanks minus the Experimental Farm. All years but 1933 had
downtown thermometers at 3 m height; 1933 was 20 m temperatures.
November monthly temperature differences between the city site and the
Experimental Farm, from which it appears that the 3 m to 20 m inversion in
the City in 1933 may have averaged around 2°C, for a near-ground inversion
strength of around 12°C/100 m. (The warmer temperatures at the Experimental
Farm are an elevation effect due to the background inversion.) This is
in sharp contrast with present day conditions because inversions are now
rare in the lowest 20 m in the city. City lapse rates are probably normal
to around 50 m and the city mixing depth is almost certainly more than
20 m. (Present-day lapse rates are discussed in detail in Section 6).
16
-------
SECTION 5
THE SURFACE TEMPERATURE FIELD
MEASUREMENTS
Information on the distribution of near-surface temperatures in the
Fairbanks area was obtained from automobile traverses, from a variable number
of thermograph stations, and from regular NOAA weather service observations at
Fairbanks International Airport (WS in Figure 2). Time distribution of
traverses and thermograph data is shown, together with data on temperature and
sky cover, in Figures 5 through 7. In addition, qualitative data on actual
surface temperatures are available from two sets of aerial thermal infrared
imagery.
Traverses
Table 3 summarizes the results of the 43 surface traverses carried out in
the course of the project. The traverse vehicle had a mast mounted on the
right front bumper and extending 2m above the ground (see Figure 8). All
traverses except 18 were carried out with a YSI thermistor probe, mounted at
2 m on the mast. The 2 m temperature was normally observed on a YSI tele-
thermometer and recorded manually together with location, time, sky con-
ditions, wind directions as indicated by vapor plumes, and any other pertinent
data. Attempts were made to record temperature with a battery-operated
millivolt recorder, but the combined problems of calibration, vibration, and
of keeping a record of location on the moving strip chart made the manual
method easier and more accurate. Temperatures were normally read to .1°C from
a scale with 1"C divisions; reading accuracy varied from about +.2°C on smooth
pavement to +.5"C or more on very rough roads.
17
-------
MOV DEC
1974
JAN
FEB MAR
1975
APR
Figure 5. Summary of weather and data collection activity, Nov. 1974 through
April 1975. From top to bottom: Double line graph gives daily
maximum and minimum temperatures. Dots in line with TR show
traverse times; x's in line with FL indicate aircraft observations.
Under wind, W's indicate multiple-observer wind observations;
single lines indicate one wind recorder in operation, double lines
indicate two or more recording anemometer records available.
Sky conditions: filled curve gives sky coverage by cloud in tenths;
open curve gives sky obscuration by ice fog. Horizontal lines:
periods of thermograph records; code letters are the same as in
figure 2.
1H
-------
20
0
0°
-20
-40
TR
FW-
1 C" -
Lr
PC"
\j r
1 1 1— IM H
i 1 . i . i
MAY JUNE JULY AUG SEPT OCT
1975
Figure 6. Same as 5 but for May, 1975 through October, 1975.
-------
20
0
-20
UF
PR
CL
-40
TR
FL
Wind
10
0
FW-
LF
CF:
SR-
i i
M I I M t-
-4 H
•II-
-l MC
-I I
-I I-
NOV DEC JAN FEB MAR APR MAY
1975 1976
Figure 7. Same as 5 but for November, 1975 through May, 1976.
-------
ce
LU
O
CJ
>-
CO
1 —
ce.
0
Oi.
CO
1—
o
5
•.
z
ex
o
O-
tt
5
LU
3:
1—
—
§
u>
IX
""••
(X
> •
u.
c>
(X
CO
CO
UJ
oo
h-
-p.
0
o
UJ
ac
LU
1—
z
z
LU
t/)
LU
ev
*—
S
LU
i
LU
t—
^—
CO
LU
_i
o
CJ
LU
a:
u.
o
1
LU
o:
"»
a-
l- -
UJ
1—
£
o;
s
o
L. S-
0) 3>
J= v>
O O
C
ID
Ol
X
h-
.
X
*
*r)
X 01
•r- O
CJ CJ
•o
C
C O
10 i.
xS1
u
(O
CO
'4^
01
f
O
^
CO
-o
C
~f
• n
O- C
El 4J
I-- 1^-
jj*
-
"*
01
O
O)
2
u
D
CU
^3-
o
<7*
l
O^ "«f r— • i/) LO O> VO
*r PO ^- o> vo r*-. ^o t»«. rx CM <\j r^ oo ^ *£> *^
LO CJ CVJ LO
cvj r**> Oil oo oo ^* ^D oo *o ^f f" *~ f*** ^* co r™~
^-cvj^foinro^j-rocM^-cgcviroro-tf^-
i i i i i i i t i i i i i i i i
ur> LO
tOi— co*jOrocvjir>Or— *ocor^.^-r-^vor*'
r— rocvj^-^-^-«j-^-fo»— csj<\j*j-co^'^'
i i i i t i i i i i i i i i i '
10
O ^1- voro mi/> otoo
OO
^r—ror^^-Lniororo^ocnoOinro^or^
r— rocvjj^^t^-^j-^j-ro»— CJCM^- i *t -«
o ±
00 O O •— O O
o^^^itt.-.It^JoS^^^'u.u.
v>
LO O I
i i ro ro
ro cj ro ** ^*.
UJ 31 UJ
3 LU 2 z to ac
CJCJCJt_)t_>CJaE»_>ZOOZUJOjBCjZ
LO
r" CD C3 r— C3 ^~ C3 ^± CO t/> CO ^— ^~ C3 ^™
+ ' + i 4 -t- +1+1 i
0
ro O
O OO CO fO
m r- CO r—
«3 f— CO lijO^J-«*r— COCsJt— CVIiOVjOr^0^
•— CM CVJ O r— »— CVJ Ir-CMCNJCSIr— ^r-CVI
iiiro 1110 iiiiiii*
too*— *a-cocvjcvjro ^-^-r^.»— ^-cvivoO
t— CMCVJf— »— r— CM Cn r- CVJ.— .-vxpN. r^-^-v^.-x^-x^rx.r-*
>^.^N.-^.rO**--«x.^1.LO1^x.'1*«^O*OOO O*"^"1*^
^- VD i— » »— * K-* X
'-•^-'•-'^>>>»-HX»— cvjfn^-LocD'-*
^^
•o'
0>
i
**
e
o
o
-------
c
0
fO
t- L-
-C i/>
+J JD
O O
C
ro
2
^
X
£
1—
•J
>, QJ
4-> 1-
•r- O
(_j CJ
C
C 0
S Jp
o
ro
CO
I/)
0)
8
CO
C
• -o
Q. C
1— I—
1
••-»
h-
CO
at
*o
0)
I/I
t—
c
CL
£
-r-
un r-*
in
in co
LO CO
in to
OJ i—
i i
in
CO 00
i i
00
cn cn
i i
r— 0
-cr cn
i i
co in
LO LO
CO ^~
"*•— ~*^-
f
c
O
t/1
L- C
ai o
c: t/>
*~ at
c
OJ •«—
oj m CM oj
ro •— co co
r-. LO r-. m
co r- cn co
co in co
in o 4-» **• LO
^" *3" (/> CO OJ
' ' 3 ' '
4-*
C
5. OJ
cn oj u co co
I I O) l l
at
fO
L.
a>
LO cn in in
a.
*j- ^ i ^- ^i
ii it
*-•
at o
X
CO
«*•
UJ
LO
Ti ° ° T T
-r | ii
in in oj cn LO
i i i i i
i — CO i — VO LO
LO LO
-s. "x. -x. O CO
LO r-. cn r~- , —
*^* *-». "v. --* — -x
*
CD i — oj ro ^-
4*
c
o
in
a>
c
LO
CO
in
OJ
LO
CO
1
o
1
s
o
**
r-
co
,
•—
in
*
"-«*
LO
C
o
I/)
at
c
LO
r«*
CO
OJ
i
LO
cn
i
o
OJ
o
o
m
CO
0
*
•—
in
i
CO
LO
OJ
OJ
*•*-
^
to
*r
CO
LO
CO
LO
OJ
1
r—
<0
1
OJ
LO
1
u
o
o
in
vo
i
CO
in
OJ
*^-
f^-
4-
a>
3
tO
L-
55
u>
in
O
OJ
i
in
CO
i
OJ
i
0
o
in
T
in
i
in
o
CO
'**-.
4t
CO
c
o
VI
ai
c
0
VO
O
VO
OJ
1
O
1
o
CO
1
CD
O
CO
in
i
O
vo
i
CO
m
in
CO
"**-
cn
c
at
c
o
w
at
c
r^
OJ
O
LO
CO
CO
OJ
1
vo
1
ro
CO
OJ
QJ
to
L.
rO
in
:x
CO
+ 1
r-
LO
OJ
cn
CO
*"-
0
0
c
1
vt
cn
c
•5
0)
j=
LT>
^
m
LO
OJ
OJ
i
r--
i
cn
OJ
i
o
o
"*
z
in
i
in
i
in
-
*^-»
^~
0
c
1
at
c
LO
•^
CO
0
LO
OJ
CO
1
O
1
CO
OJ
1
u_
I/I
JZ
4-*
rO
o_
oo
LO
+
o
CO
o
CO
LO
2
"*•*-.
t_J
OJ
d.
0
•r-
'&
OJ
vo
OJ
0
CO
CO
1
o
i
8
1
o
o
Tl
LO
l~^_
o
1
CO
CO
CO ^~
^- **s.
CO
•f-
-------
•o
c
o
.r- 01
"o> c
U 01 ••••
1 01 £
in i—
•— oo
ou r-~
u> in
oo
ro
ro
%
in i—
r— r- CSJ
^ o •—
in in o o CM i—
i i i t i i
o
o
o
\
o
o
-^
o
i
--v. U_ U_
CM •-. «-«
«0
O O O O O
^ o fe ^ ^ ^*
(_}(_> C_> O
O t—, r-, •—
+ 1 +1
i— r- CM i— f~
&
CM r— .—
s
>0 r— t—
S
CO CO
23
-------
a.
b.
T,
'2-J
OEEEO
isr
Figure 8. Locations of thermistor sensors on traverse vehicle, (a) side
view; (b) front view; (c) detail of thermistor sheild (used for
late spring and summer daytime runs.)
With the exception of daylight runs more than two months from the
winter solstice, the thermistor was shielded only on an experimental basis.
Shielded and unshielded thermistors operated side by side on a sunny February
day indicated no difference in equilibrium temperature while the vehicle was
moving, but a decided slowing of response in the shielded thermistor after
stops. As only temperatures obtained while the traverse vehicle was in
motion were being recorded, the shielding was considered a disadvantage.
Ground inversion strength was recorded by adding a second thermistor of
the same type .5 m above the ground and approximately 30 to 40 cm in front
of the mast (Figure 8). Traverses with this two-thermistor arrangement are
marked "inversion" in Table 3. During the initial test run, 12, the second
sensor was mounted on the bumper, but with this arrangement the engine
arid/or headlights of the traverse vehicle appeared to be heating the lower
sensor. Inversion strength was recorded on a strip chart as the difference
in voltage output from the two telethermometers. As the calibration of this
arrangement was dependent on the temperature, an approximate calibration
technique was used. Periodically during a run the observer would note the
temperature difference between the two telethermometer dials and mark this
24
-------
on the strip chart. Location was also marked frequently. As the electronics
were temperature sensitive, calibration also involved repeated re-zeroing of
the telethermometers and (usually simultaneously) checking the zero on the
portable recorder. The thermistor sensors were occasionally run side by
side to confirm that they were in agreement with each other.
The noise level in these near-ground inversion strength records was
quite high due to the fact that mechanical vibration of the glavanometer
needles in the telethermometers added a spurious electrical signal to the
recorder input -- and mechanical vibration on some of the roads used for
traverses was considerable. Consequently, the .5 m to 2 m temperature
difference normally had to be of the order of ,5°C (a lapse rate or inversion
of 33 times the adiabatic lapse rate) in order to be detected. Temperature
differences considerably in excess of this amount were in fact detected
frequently, as will be discussed below.
Attempts were made to observe the incoming infrared flux, using a
pyrgeometer mounted on the roof of the traverse vehicle. Although the
records obtained appeared to be of good quality, especially in comparison
with those from the thermistors, attempts to relate them to physical reality
suggested that the instrument was unreliable under conditions of rapid
temperature change. Tests on a building roof confirmed that the pyrgeometer
was in fact highly sensitive to heating of the (supposedly) infrared-transpar-
ent dome. As our primary interest was in comparing the incoming radiation
at ground level in the warm, foggy city with that in the cold outlying
areas, the pyrgeometer data were unsuitable. However the pyrgeometer
demonstrated a significant increase in incoming radiation whenever the
traverse vehicle passed under a visible stack plume from a power plant.
Traverses were heavily used in the winter months, when the diurnal
temperature variation was small and the temperature curve was quite flat
over at least part of the day. During the summer months the diurnal temper-
ature cycle was far stronger, and periods of near-equilibrium temperature
did not occur. No effort was made to run traverses under these conditions,
as the differences in cooling and heating rates between city and background
would have made interpretation of the results extremely difficult.
-------
Thermographs
A thermograph network was maintained throughout the project, although
the number of active instruments varied considerably. Each thermograph used
was calibrated at least twice against the thermistor used for the traverses
at 5°C to 10°C intervals from 0° C down to -45°C with the aid of a low-
temperature freezer chest. (In most cases the full calibration was not
possible, as the thermograph clocks and/or inking systems quit between -35°C
and -40°C). At temperatures above 0°C, calibration was normally carried out
in the open, with the thermistor as near as possible to the thermograph
sensor element. Allowance was made for the poor ventilation of the thermistor
sensor, and in any case the intercomparison among thermographs should not be
affected by this. Those thermographs which required range changes between
summer and winter were calibrated before and after each change.
Thermograph sites are shown by the circled pairs of letters on Figure 2
and the period of operation at each site is summarized in Figures 5-7. Gaps
in the records are due to: (1) non-operation or illegible records at tempera-
tures below -30°C; (2) inability to reach some sites under some weather
conditions; (3) removal of thermographs for calibration and servicing; (4)
thermographs being required for other projects; (5) one or more thermographs
simply not working; (6) vandalism or wind damage to the shelters and (7) at
Peger Road, interference from a sewer construction project. Whenever possible,
we attempted to maintain records from City Lot (CL), Creamers Field (CF) and
Peger Road (PR), in that order of priority, at the expense of the other
stations if necessary.
Thermograph CL--
The maximum city temperature observed on traverses was generally located
at 2nd Avenue and Cushman Street (Fig. 9). Thermograph site CL (City Lot)
was located approximately 200 m south and the same distance west of this
maximum, at the west end of a parking lot. Figure 10 shows the site as it
appeared in spring 1976. Several traverses included a loop through this
parking lot, after which the temperature recorded from the car was compared
with that of the thermograph. At this location, thermograph and traverse
26
-------
temperatures were generally equal and, under strong heat island conditions,
about 2°C lower than the temperature at 2nd and Cushman.
Figure 9. Vertical view of downtown Fairbanks, showing the heat island area
(white arrow) and the thermograph site CL.
27
-------
Figure 10. Thermograph site CL, looking south-southeast, in late April 1975.
temperatures were generally equal and, under strong heat island conditions,
about 2°C lower than the temperature at 2nd and Cushman.
Thermograph CF--
Site CF (Creamers Field) was located at the site of a former dairy
farm. Much of the area is now a wildlife refuge, primarily for northward-
migrating waterfowl. Figure 11 shows the site as seen from the air; the
peculiar grid pattern southeast of the thermograph site represents a snow
clearance effort for the benefit of the wild geese which normally come
through in late April. (This photo, along with Figures 9 and 13, was taken
near noon on April 21, 1975). As can be seen from Figure 12, the area is
quite flat, but not actually swampy except during breakup season; it is
regularly sown to grain crops. This thermograph site and the road leading to
it provided about half of the minimum temperatures in Table 3. As discussed
in Section 4, this site was the closest we could come to a good background
site "upwind" of town.
-------
Figure 11
Vertical view of Creamer's Field, College
showing thermograph site CF, 21 April 1975
Field, College Road at lower edge,
April 1975.
29
-------
Figure 12. Thermograph site CF, looking NNE, April, 1975.
Thermograph PR--
Site PR (Peger Road) was initially located in natural muskeg vegetation
near the north bank of the Tanana River (Figure 13). It was subject both to
air drift from Fairbanks (as evidenced by ice fog) and to considerable site
disruption. The shelter was initially installed in November, 1974. When
the snow melted in April, it became apparent that a low area between the
shelter and the road was actually a stagnant slough of the Tanana River.
The area which appears grassy in Figure 14 was knee-deep or more in water
through most of May and June. At roughly 10:00 p.m. on May 20th, an unknown
person put a bullet through the thermograph drum, apparently while using the
shelter (which had about 25 bullet holes in it) for target practice. The
thermograph was replaced and serviced through July with the aid of hip
boots, but of course the second calibration on the damaged instrument was
lost. In August, excavation began for the new Fairbanks sewer outfall, and
a water-filled ditch between the road and the shelter made access impossible
until the pipe was installed and the ditch backfilled. We wanted a wind
station in the area, so since the vegetation around the original site was
thoroughly disrupted anyway, the site was moved about 100 m north to a flat
30
-------
area where both winds and temperature could be measured. Figure 15, taken
from almost the same point as Figure 14, shows the new site as well as the
two tall spruces (on the left) which had marked the original site. Site
installation was further delayed by a bulldozer knocking down the anemometer
support post (which was supposedly well out of the construction area). The
thermograph was finally installed in early November, and wind records were
started in December. The station was removed because of rising water in
late April. Figure 16 was taken while the station was being dismantled.
Figure 13. Vertical view of Peger Road area, 21 April 1975, Tanana River at
lower part of photo. Peger Road thermograph locations marked by
PR
31
-------
Figure 14. Original site PR looking west from access road. Summer
1975.
32
-------
Figure 15. (Above) Both PR sites, photo taken April 1976 from the same loca-
tion as was Figure 14. The tallest spruces at the left mark the
old site; the thermograph shelter is at the new site.
Figure 16. (Below) Both PR sites, looking SE, anemometer at the new site in
the foreground.
33
-------
Thermal Infrared Imagery
Two sets of thermal infrared imagery of the Fairbanks area were available
to the project. One was flown at about 5 am Alaska time March 4, 1975 by
the 172nd MID(AS) operating out of Fort Wainwright, Alaska. Temperatures at
2 m elevation were approximately -14 to -24°C. Surface and helicopter
measurements had been carried out between 23:30 and 01:30 the same night.
Figure 17 shows the temperatures recorded by the thermograph network that
night.
The second set of thermal imagery was flown sometime between July, 1969
and June, 1971, but we have not been able to trace the exact date or the
organization responsible. On internal evidence it was flown in summer,
probably at night in summer 1970.
o
o
uf
o:
I
DC.
Ul
£L
2
UJ
-20 -
-30 -
16 20 00 4 8 12
MARCH 3 MARCH 4,1975
LOCAL TIME
Figure 17. Thermograph temperatures, night of March 3-4, 1975 (23). Stations
keyed to Figure 2.
34
-------
The thermal infrared images provide excellent qualitative information
on surface temperature variations, and thus on important heat sources for
conductive/convective heating of the air. They do not normally provide
information on heating of the air due to direct admixture with the ambient
air of hot gases escaping from heated buildings or produced by combustion
processes.
RESULTS
The simplest heat island to explain from a theoretical point of view
was that observed under near steady-state conditions in mid winter. This
type of heat island will be described first, followed by a discussion of how
the heat island intensity changes with season and time of day. Finally the
pattern of near-ground inversion strength will be considered.
Heat Island
The winter heat island is of considerable interest both theoretically
and because of practical observational aspects. The heat island at this
time of year is due primarily to direct anthropogenic heating and effects of
air pollution, with possible secondary effects from surface modification
(packing and removal of snow). The diurnal thermal cycle is negligible from
late November to the end of January and the nighttime temperature shows a
distinct plateau (a period of several hours of fairly steady temperatures)
from October through March. From a practical point of view, this allowed
traverses to be carried out and interpreted without a great deal of correction
for cooling or heating during the traverse. Also the plateau in temperature
suggests a steady state thermal regime, and thus a steady state heat island.
Most of the traverse data listed in Table 3 were obtained under these
plateau temperature conditions, the exceptions being 16A, 17, 19, 22A, 25, 26,
28, 35, 36, 38, and 39. Four of the winter cases, plus two in which some
adjustment of temperature was necessary, are described below. These six
cases were selected on the basis of completeness of coverage (four were run
simultaneously with aerial observations, one had the best ground inversion
data in ice fog, and one had simultaneous wind measurements) and variability
35
-------
in the degree of ice fog and cloud coverage.
Heavy Ice Fog; Traverse 8--
Figure 18 shows the pattern of temperatures around Fairbanks from 1300-
1400 January 3, 1975. Ice fog was intense, with airport visibilities 1/4
mile during the traverse and dropping to 1/8 mile later in the day; hori-
zontal visibility downtown was considerably less. Figure 19 gives a good
impression of the density of the ice fog, which looked even thicker from the
ground. The contrast between only slightly impeded vertical visibility and
very poor slant or horizontal visibility is typical of ice fog.
Figure 18. Temperature contours at 2m elevation for a typical heavy ice-fog
situation, 1300-1400 January 3, 1975 (8). Contours labeled in °C.
36
-------
Figure 19. Aerial view of Fairbanks during traverse 8, 3 January 1975.
A light airplane with thermistor was flown during this traverse from an
airstrip outside of the foggy area; the temperature measurements aloft are
discussed in Section 6. Measurements are predominantly from above 200 m, as
horizontal visibility at 200 m was too low for safe flying. Figures 19 and
40 were taken from this plane.
The temperature pattern is fairly typical of those observed with heavy
ice fog. Thermal gradients were relatively weak except at the Chena River
bridge north of town and along the hill slope by the University, and even in
these two places they were weak compared with the clear-sky traverses, 24
and 38. The heat island intensity was about 5.5°C. Most likely this
smoothing of horizontal temperature gradients is due to the reduction of
vertical temperature gradients, as the "background" areas were generally
fogged in during severe ice fog (see Section 9). If the temperature along
the slope in front of the Geophysical Institute at a height of 60 m above
the flats is taken to be an approximation of the 60 m free-air temperature,
and Figure 35 (Section 6) is used as a guide to the relationship between
heat island intensity and inversion strength, the ice-fog heat islands are
within the expected range except for traverses 4, 6 and 7, in which the city
heat islands are larger than would be predicted.
37
-------
Even though the fog is not optically thick to a vertical beam, cal-
culations using observed crystal size distributions and densities show that
radiative transfer in the thermal infrared is probably sufficient to de-
stabilize dense fogs, or even moderately dense fogs when anthropogenic
heating is considered (Bowling, 1970). Weakened or missing inversions in
the first 100 in of the atmosphere are not always visible on standard Weather
Service Rawinsonde records, but occasional cases of normal lapse rates
within layers of dense ice fog have been observed for over 20 years (Robinson
and Bell, 1956).
Moderate Ice Fog; Traverses 13 and 29—
According to NOAA Weather Service observations, visibility at the
airport was about 1 mile during traverse 29 (which was run simultaneously
with a helicopter flight,) and 1/2 to 1/4 mile during traverse 13. Isotherms
at 2 m elevation are shown in Figures 20 and 21. Horizontal temperature
Figure 20. Temperature contours at 2m elevation for Traverse 13, 1600-1900
January 10, 1975. Moderate ice-fog.
-------
Figure 21. Temperature contours at 2m elevation for Traverse 29, 2000-2200
December 1, 1975. Moderate ice-fog.
gradients were strongest in traverse 13, and those in traverses 8 and 29
were rather similar in appearance. The shape of the heat island was similar
in all three cases - a sharp maximum near Second Avenue and Cushman Street
with a strong gradient just to the north at the Chena River and a more
diffuse maximum to the south along Cushman Street and west along Airport
Road.
Although the heat island intensity does not correlate well with ice fog
density in these three cases, it does correlate well with inversion strength,
as can be seen from Figure 22. The reason for the discrepancy between
inversion strength and ice fog density may be found in the way an ice fog
episode develops (Bowling et al., 1968). During the early stages (January 2
and 3, 1975 in this case) cold air advection aloft and radiative cooling
39
-------
Figure 22. Comparison of low-level sounding data for times near Traverse 8,
(1/3/76), 11 (1/7/75), 13 (1/10/75), all based on rawinsonde data,
and traverse 29 (12/1/75), based on helicopter data. Traverse 11
had the weakest nighttime heat island in the study.
1000
900
800
700
600
500
4OO
300
200
100
3 Jan !975, 1400 AST
— - 7 Jan 1975, 1400 AST
-•- 10 Jan 1975, 1400 AST
•••• I Dec 1975, 2100 AST
-40
-30
Temperature,°C
-20
-10
-------
from the ground act together to lower the temperature of the entire air
column. Warm air advection then takes over aloft but if sufficient ice fog
is present to affect radiative cooling from the ground, the ground inversion
may remain constant or weaken even while the overall inversion in the first
kilometer strengthens. Three days of ground temperature below -45°C allowed
buildup of enough ice fog to produce this situation by January 7. Near the
end of an ice fog episode, warm air advection aloft strengthens markedly and
extends to lower levels, resulting in the strength of the ground inversion
being increased by heating from above. As this heating also erodes away the
top of the ice fog, radiative cooling from the ground can also strengthen at
this time — thus the steeper inversion and stronger heat island while the
ground level fog remains quite dense near the end of an ice fog episode.
Clear Skies; Traverses 24 and 38—
Winter and early spring nights (and December and early January days)
with clear skies had the most intense temperature contrasts observed.
Figures 23 and 24 show the results of two winter-night traverses. One
feature of the clear-sky heat island which is immediately obvious in Figure
23 is the extreme variability of background temperature. Some features, such
as the low temperatures and occasional sharp thermal gradients along College
Road (northwest of the city core), are fairly consistant among traverses and
probably reflect the gravity-driven micro and/or mesoscale circulation
patterns of the area (see Section 7). The higher temperatures southeast and
south of the city in Figure 23, however, are examples of "random" warm spots
which occasionally appear in clear-sky night-time traverses. It should be
noted that these warm spots represent a weakening but not any actual breaking
of the thermal inversion - temperatures measured 100 m above the surface with
a helicopter during the time covered in Figure 23 ranged from -3°C north of
town to-5°C over the Tanana River (see Figure 39). The warm spots may be due
to partial vertical mixing by local winds or to a combination of anthropogenic
sources and local winds. Wind data were not available.
Returning to the more persistent features, essentially all the clear-sky
nighttime traverses had primary temperature maxima at Second Avenue and
Cushman Street and in front of the airport terminal building, with the
41
-------
Figure 23. Temperature contours at 2m elevation for Traverse 24, 2320
13-0120 14 March, 1975. Clear skies.
downtown area generally showing a secondary maximum about a kilometer farther
south, at the intersection of Cushman Street and Airport Road. (The warm
area at the University, WNW of town, is due to the University being on a
hill.) The lowest temperatures were normally found in isolated patches
along College Road, often bounded by very sharp thermal gradients. These
cold areas were interpreted as tongues of cold air flowing across College
Road from the Creamer's Field area. Those traverses which extended west-
southwest of the University (e.g., 38) frequently encountered an equally
cold or even colder area, suggesting that these very low temperatures may
have been realistic background temperatures. Temperatures south of town
were generally lower than those in town, but not as low as those to the
north. Figure 24, which shows the available wind directions as well as the
42
-------
Figure 24. Temperature contours at 2m elevation for Traverse 38, 2100-2230
February 26, 1976. Clear skies. Measured wind directions shown
by heavy arrows; wind directions inferred from plume drift shown
by light solid arrows (ground level) or light dashed arrows (high
stack plumes).
temperature field, suggests that the difference could be due to the fact
that air encountered south of town had mostly been heated as it moved through
the built-up area and was cooling again.
Cloud Cover; Traverse 17--
During the five traverses described so far, skies as seen from hills in
the area were clear. Traverse 17, however, was carried out under a complete
overcast, with ceiling at about 12,000 feet (3500 m). Figure 25 shows the
resulting temperatures and general wind directions. The 60 m inversion as
measured between the Geophysical Institute and the coldest background
station was only 3°C, and the heat island was correspondingly weak. Even
43
-------
Figure 25. Temperature contours at 2m elevation for Traverse 17, 1430-1600
January 25, 1975. Overcast skies. Arrows show approximate mean
wind directions over the half hour period 1440-1510.
the gradient at the Chena river crossing north of town was weak, possibly
because the downtown winds in this case were observed to be primarily from
the south. (In agreement with this, the lowest temperature recorded was
found on south Peger Road rather than in Creamer's Field).
Only a few traverses were carried out under fully overcast conditions,
but 17 seems fairly typical. Overall heat island intensities were from 2°C
to 3.5°C, and background temperatures north and south of town were similar.
As the usual air drift from the north in the area is caused by radiative
cooling on the hills to the north, the cloud cover, which would have reduced
the radiative cooling substantially, may be directly responsible not only
for the weak thermal gradients, but for the relatively uniform background
temperatures as well.
44
-------
Seasonal Variations
Seasonal variations in the intensity of the heat island can be traced
both to surface changes (e.g., dry snow, wet snow, bare ground, dormant or
growing vegetation) and to major changes in the pattern of insolation. The
effect of changes in the day length and solar elevation angle during the
period of continuous dry snow cover will be considered first, followed by
the influence of breakup (snow melt) and the variation in solar radiation
during the period when the ground is snow-free.
The Dry-Snow Season--
Thawing temperatures and even rain are possible in any month of the
year in Fairbanks, but are highly unusual from late October through early
March. The months of November through February and sometimes March are
characterized by a complete cover of dry snow which is an excellent in-
sulator in the outlying areas, but is normally partially removed and the
remainder compacted in the city area. Traverses 1-23 and 28-38 were carried
out under these conditions. There is some increase in snow depth through
each period.
The solar radiation regime shows marked changes through the season of
dry snow. Table 4 shows that day length in the November-February period
varies from 3 hours 42 minutes to about 10 hours, with maximum solar elevation
(approximately the angle above the horizon of the sun at noon) varying from
1°44' to around 16°. Since the intensity of sunlight on a horizontal surface
is proportional to the sine of the elevation angle (neglecting the effect of
the atmosphere) the intensity of the noon-hour solar radiation varies by
almost an order of magnitude. When the doubling of the day length is also
taken into account, it is not surprising that the possible daily total solar
radiation varies by about two orders of magnitude.
During December and much of January, there is so little incoming solar
radiation that the daily thermal cycle is lost in random temperature changes.
Temperature changes are due to changes in the air (including cloudiness)
several thousand meters up or to changes in the efficiency of wind mixing
rather than to any regular daily changes, and if these upper air conditions
stay constant for several days a near-steady state may develop at the ground.
Under these circumstances, neither the albedo nor the thermal properties of
45
-------
TABLE 4. DAY LENGTH AND MAXIMUM SOLAR ELEVATION AT FAIRBANKS (64°50'N) AS A
FUNCTION OF TIME OF YEAR. ALL TIMES BASED ON 150°W MERIDIAN TIME
(ALASKA STANDARD TIME).
Date
Jan. 21
Feb. 21
March 21
April 21
May 21
June 21
July 21
Aug. 21
Sept. 21
Oct. 21
Nov. 21
Dec. 21
Sunrise
Time
9:11
7:30
5:49
3:55
2:09
0:59
2:13
3:56
5:31
7:02
8:47
9:59
Sunset
Time
14:55
16:42
18:10
19:47
21:30
22:48
21:40
19:50
17:57
16:09
14:27
13:41
Length
of Day
5:44
9:12
12:21
15:52
19:21
21:49
19:27
15:54
12:26
9:07
5:40
3:42
Maximum
Solar Elevation
5°05'
14°20'
25°10'
36°45'
45°12'
48°37'
45°48'
37°33'
25°10'
14°45'
5°25'
1°44'
the surface are of major importance except during a rapid change from one
type of weather to another. Earlier and later in the winter season, however,
the diurnal cycle is well developed and the effects of the removal, dis-
coloration and compaction of snow become quite important.
As a very rough first-order approximation, suppose we approximate the
daily variation in the radiation balance by a sinusoidal variation in the
heat conducted through the surface of the substrate. In this case, it is not
too difficult to show that if the energy flux cycle has the same amplitude in
all substrates, the amplitude of the thermal cycle will vary as the inverse
square root of the product of the conductivity, the density and the specific
46
-------
heat (l//cplc ). Table 5 lists these properties for various building
materials and for snow of various degrees of compaction. It should be kept
in mind that the actual temperature variation will depend on albedo as well
as thermal properties and variation of the incoming radiation. The values in
the table are probably most useful in estimating relative cooling rates
immediately after sunset, or after a sudden change in the incoming thermal
infrared radiation. It is interesting to note that dry snow packed (and
dirtied) by vehicular traffic has thermal properties surprisingly similar to
those of concrete or asphalt, while a fresh snow surface is capable of ex-
tremely rapid response to changing radiative conditions.
Two sets of traverses were carried out specifically to obtain infor-
mation on the dependence of the heat island on the time of day. Traverse sets
16A, B and C and 22A, B and C each include one traverse in the warmest part
of the day, one a short time after sunset and one just before sunrise. The
cooling rates between the early-afternoon and late-afternoon traverses of set
22 are particularly impressive, with the Creamer's Field temperature dropping
by 14°C. while the downtown area cooled by only 5°C. The traverse vehicle
was always on asphalt or firmly packed snow, although the surroundings of the
roads varied greatly, so the difference between the city and background
cooling rates is probably greater than that recorded.
The importance of the substrate at this time of year is clearly evident
in the thermal infrared imagry of Figures 26 and 27. Both figures are from
the March 4 flight. The lightest and warmest areas (other than open water
areas such as the Chena River) are heavily travelled roads where the surface
is asphalt or asphalt covered with clear ice. Parking lots and lightly
travelled roads generally had a veneer of packed snow, giving them a much
higher albedo. This leads to a lower daytime temperature which in turn is
reflected in lower 5:00 a.m. temperatures. Figure 26 shows some particularly
interesting effects, such as the "scribbled" looking area on the left side of
the large parking lot below the round building in the upper center. This
appearance, which is common in accessible open areas in the vicinity of the
city, can only be attributed to snow machine tracks. Albedo is little
affected by a single snow machine pass, and these tracks, which are con-
spicuous in infrared images from 1000 feet elevation, would not be detectable
in visible light. Finally, the low-level inversion itself is visible in the
47
-------
TABLE 5. THERMAL PROPERTIES OF VARIOUS SUBSTRATES FOUND
IN THE FAIRBANKS AREA. (SNOW AT -20°C.) l//c^lT IS
PROPORTIONAL TO THE AMPLITUDE OF THE THERMAL CYCLE WITH A
GIVEN ENERGY-BALANCE CYCLE
Substance
moist sandy clay
brick masonry
concrete
asphalt
Density, p
g cm"
1.78
1.7
2.3
2.2
Specific
Heat, c
cal gnf1 °C~]
.33
.20
.20
.20
Thermal
Conductivity, k
cal sec"1 crn"1 ''C"1
.0022
.0015
.0022
.0036
V
/cpk"
28
44
31
25
snow, freshly fallen .1
snow, average
undisturbed .2
snow machine track .4
snow, packed by auto
traffic .8
.468
.468
.468
.468
.00018
.00036
.00097
.00400
345
172
73
26
mottled or patchy bright areas at either end of the parking lot. The bright
spots are tree tops which are in higher and warmer air.
Figure 27 shows several thermal features related to heat release due to
human activity directly rather than through surface modification. The most
notable of these is the Chena River, which under natural conditions, as is
visible in other images upstream from any addition of waste water, would be
the coldest area in the image. Instead, waste water from Fort Wainwright,
added upstream from this image, maintains a warm channel through the area
crossed by traverses. Cooling water dumped by the Municipal Utilities System
(MUS) power plant at the point where the entire river becomes a conspicuous
bright band keeps the river well above natural temperatures for several
kilometers. There are no bridges in the warmest area, but the effect of the
river on air temperatures on the downwind bank must be considerable --
43
-------
* *ft
Figure 26. Thermal infrared imagery of the Alaskaland (AL) area, 0500 March 4,
1975. The general gradient of density from light at the bottom to
dark at the top is an artifact of processing and should be ignored
(Photo courtesy of U.S. Army 172nd MID(AS) and Cold Regions
Research and Engineering Laboratory).
Figure 27. Thermal infrared image of part of downtown Fairbanks, about 5 a.m.
on a clear March morning. Note the importance of the river as a
heat source, especially downstream of the Municipal Utilities
System power plant near the center of the photo. The general
gradient of density from light at the bottom to dark at the top is
an artifact of processing and should be ignored (photo courtesy of
U.S. Army 172nd MID(AS) and Cold Regions Research and Engineering
Laboratory).
49
-------
especially when ambient temperatures are in the -40°C. range and the river is
unfrozen. (The effect of the open water on ice fog in the same area has been
discussed by Benson (1970) and by Ohtake (1970).) The general warmth of the
city core is indicated by the bright haze just south of the Chena River
bridge. The bright line extending south and then southeast from the MUS
plant to the city core is due to a buried steam line. Thus the infrared
images document both the increased thermal lag of disturbed surfaces and the
direct addition of heat.
In summary, the daytime heat island during the dry-snow season becomes
progressively weaker as the insolation increases. On the basis of thermograph
records the mid-day heat island is essentially gone by late March. The
nighttime heat island changes very little. The greatest heat islands
observed occur in the evening hours, late in the dry-snow season or in the
earliest part of the melting season (Table 3).
Snowmelt--
The snowmelt season is defined as the period when water is present in
both liquid and solid phases in sufficient quantity that the latent heats of
freezing and thawing are a non-negligible part of the energy balance. In
Fairbanks this normally includes most of April, at least the latter part of
March, and on occasion a few days in early May or late February.
During the earliest stages of snowmelt, air temperatures generally
remain below freezing and solar radiation is of primary importance. Melting
in undisturbed areas is confined to the immediate vicinity of south-facing,
near-vertical dark surfaces -- e.g., the trunks of isolated trees. South-
facing berms or cuts left by snow-removal equipment and darkened by vehicle
exhaust or sanding operations are also subject to very early melt, as is snow
in some locations on buildings. There is little or no runoff at this stage,
and water is present in the liquid phase during the warmest part of the day
only. The primary effect of this preliminary melting is to decrease the
albedo of the snow surface still further, especially in areas where the snow
is dirty.
Melting in flat areas occurs first in areas of dark ice and heavily
soiled compacted snow -- i.e., on main roads. Reflected sunlight and reradi-
ated thermal energy from buildings may accelerate the melting process in
town, with the result that the downtown area has a considerable amount of
50
-------
running water in the daytime hours while snow — even on roads —away from
the city remains dry. The latent heat of freezing in the evening causes an
effective increase in the specific heat which retards cooling even more than
in the dry-snow season. Maximum daily 2 m air temperatures at this stage of
melting may range from slightly below to slightly above freezing. Figure 28
shows an example of daily temperature variations early in this part of the
melting season, while daily minimum temperatures at several locations are
compared in Figure 29.
o
(Cloud Cover 0 Throughout)
13 14
DAY, MARCH 1975
Figure 28. An example of daily temperature variations in town (solid line,
thermograph CL) and at Creamer's Field (dashed line
It may seem puzzling that air temperatures can exceed freezing in the
outlying areas without the snow surface temperature reaching the melting
point. How is solar energy being transferred to the air? The widespread
occurrence of trees, especially of low-albedo spruces, is probably the
explanation. Solar energy is transferred to the air via trees at heights
somewhat higher than the snow surface and an inversion is maintained very
close to the ground. Advection of above-freezing air may be involved later
51
-------
in the season, but seems unlikely in itself to account for the pronounced
daily peaks in Figure 28.
MARCH minimum T 1975
-40
10
14 18
MARCH 1975
22 26
30
Figure 29. Comparison of daily minimum temperatures at four locations in
March. Heavy solid line is site CL (downtown), light solid line is
the official weather service temperature recorded at the airport,
(WS), dashed line is Creamer's Field (CF) and dotted line is Peger
Road (PR), both outlying stations. Squares show sky conditions,
with open squares for clear skies, black squares for complete over-
cast and intermediate shading for intermediate conditions. Slant
of shading indicates increasing (up to right) decreasing (down to
right) or constant (crossed lines) cloudiness.
52
-------
The major features of interest in Figure 28 are (1) the similarity of
daytime temperatures in city and country; (2) the difference between city and
country in evening cooling rates, and (3) the amplitude of the background
temperature cycle -- as much as 27°C on the 13th and 14th. Note that traverse
24 (Figure 23) was carried out that night.
Snowmelt is a positive-feedback process even in undisturbed areas. The
albedo of snow decreases when it becomes wet. Also, dirt in the snow remains
as a concentrate at the snow surface, further lowering the albedo. This
feedback process operates most strongly in city areas where the snowpack is
dirtiest. Thus the downtown area becomes free of snow long before the out-
lying areas for a variety of reasons. The heat island and the dirtiness of
the snow cause an earlier start to snowmelt, the contained dirt makes the
feedback process more effective, and snow removal in the winter leaves less
snow to be melted in the spring.
Given the right weather conditions, the city core could become snow-free
while the outlying areas still had dry snow. More often, there is at least a
short period with wet snow in both areas, followed by a period when the city
is snow-free but the outlying areas still have a substantial snow cover.
(Figure 30 and most of the vertical-incidence photographs previously pre-
sented). Note that at this time the albedo of the city is approximately that
of deciduous forest - less than that of snow-covered fields, but greater than
that of dense spruce forest. Figure 31 shows how the daily temperatures
varied a few days after the photograph in Figure 30 was taken. In comparing
this figure with Figure 28, it is apparent that both the extreme daily
temperature variations and the contrast between the city and country cooling
rates have diminished greatly. Furthermore, the presence of wet snow in the
outlying areas but not in town has resulted in the daytime temperatures again
being higher in the city. This difference again disappears as the snow cover
melts fully.
The Summer Season--
Summer is here defined as the period between the completion of snowmelt
and the re-establishment of the snow cover in fall. True darkness at night
occurs at the very beginning of this time period (unless snowmelt is con-
cluded very late) and from mid-August onward. The days are normally shorter
than the nights by the time the snow cover is established in October. Once
53
-------
Figure 30. Oblique aerial photograph of Fairbanks, Alaska, looking north,
taken during the snowmelt season, April 21, 197rs. Note the lack
of snow on roads and in the city core area just south of the
wider bridge (Cushman Street). For comparison with Figure 2, the
broad E-W highway near the bottom of the figure is Airport Road
and the road bordering the snowcovered field at the top is
College Road. The Creamer's Field thermograph is at the end of
the side road north near the west edge of the subdivision roads
south of College Road.
-------
10
0
u
-------
35 r
10/10
cloud0
5 6
Day, July 1975
10
Figure 32. Temperatures and cloud cover during a relatively clear period in
early July, 1975. Stations are the same as Figure 29.
Figure 32 shows an example of summer temperature variation during the
time period when the sun drops only a few degrees below the horizon. The
insolation at this time is very close to being a full sine wave, and the
temperature wave follows this sine curve rather closely. The amplitude of
the summer temperature wave is somewhat less than it was in March, even
though the amplitude of the variation in incoming radiation is much larger;
this is probably due to the change in substrate from snow to grass. The fact
that the night is too short to allow even an approach to the radiative
equilibrium temperature may also play a part.
A set of thermal infrared imagry flown on a summer night in 1970 (Figure
33) shows several points of interest including the following: 1. The depen-
dence of nighttime temperature on very slight differences in elevation is
clearly visible in Creamer's Field, at the top of the picture. The fact that
the lowest-lying areas show up as bright spots within the darker swales
suggest that stagnant, very shallow water was present in these areas. Such
-------
Figure 33. Thermal infrared
imagery of the
Fairbanks area in
summer. Site CF is at
the top edge. This
scene was dated to
approximately summer
1970 largely on the
state of completion
of the intersection of
Airport Road and
Cushman Street, near
the lower edge.
-------
water would be relatively cool by day, but would retain the day's heating
longer at night than would the land surface. Deeper water, as in the Chena
river and the gravel pit just below the first residential area, is colder
than the land even at night. 2. The downtown area appears warmer than its
surroundings primarily because of the very large fraction of area covered by
artificial surfaces. In the original, it is possible to discern the pattern
in which cars were parked in parking lots during the day, as these areas of
pavement were shaded and were not heated to the same extent as were the
traffic lanes. 3. The influence of the power plant on the Chena river,
although still visible where the river runs out of the picture to the left,
is vastly reduced from its effect in winter.
Grpjund_Jjiyersjpn Strength--
A number of traverses were carried out with thermistors mounted at .5 m
as well as 2 m. The resulting data on near-ground inversion strength gave
rise to the following generalizations:
1. Visible exhaust plumes were almost invariably associated with strong
temperature gradients. When cars passed the traverse vehicle from either
direction and their visible plumes reached the right front corner of the
traverse vehicle, the result was an immediate shift toward less stable con-
ditions - changes of 1°C/1.5 m (67°C/100 m) were normal. Exhaust plumes from
stationary vehicles produced superadiabatic lapse rates if the plumes stayed
low (e.q., if a light wind were present) or produced apparent inversions if
the exhaust plumes were rising. It appears that the heat contained in an
auto exhaust plume is sufficient to insure mixing through at least the first
2 m of the atmosphere.
2. On streets with heavy traffic and ice fog, but with the traverse
vehicle not being obviously affected by a distinct plume, the vertical
gradient was zero with'in the limits of our measurements. (Isothermal and
adiabatic conditions could not have been distinguished from each other).
3. On streets with moderate to heavy ice fog but without appreciable
traffic, inversions of as much as .5°C/1.5 m (33°C/100 m) in town or almost
1°C/1.5 m (67°C/100 m) out of town were observed.
b8
-------
4. On one occasion the car crossed a heavily traveled road (Airport
Road) while itself running on a very lightly traveled road (Peger Road) with
fog conditions approximately constant. An inversion of approximately 1°C/1.5
m (67°C/100 m) on the lightly traveled road changed immediately to a weak
normal lapse across the heavily traveled road; in addition, the 2 m tempera-
ture increased by 2°C. If the lapse rates on and off of Airport Road were
assumed constant up to the intersection temperature, the temperature distur-
bance due to traffic would have extended to a height of 5m; in fact, both
the normal lapse and the inverson probably decreased in intensity with
height, in which case the disturbance extended to some height above 5 m.
5. Away from the fog, an inversion of 1°C/1.5 m (67°C/100 m) up to an
extreme of 3°C/1.5 m (200°C/100 m) was normally observed. Maximum inversions
were found on gentle slopes or extended flat areas; low temperature hollows
(e.g., Ballaine Lake) were more nearly isothermal. Presumably air drainage
produced isothermal cold pools in hollows. Slope inversions may reflect very
shallow cold-air drainage in progress.
6. The maximum normal lapse rate observed from the moving car was
approximately 1°C/1.5 m (67°C/100 m) in the downtown core area in heavy
traffic. When the car was stationary with lights off, values as high as
6°C/1.5 m (400 times adiabatic) were observed. Since this was a night run,
the thermistors were probably giving a reasonably accurate picture of the
temperature distribution in the immediate vicinity of an idling car,
especially under heavy ice fog conditions when thermal radiation from the
engine compartment would be transferred via the suspended ice crystals to the
air.
7. Horizontal temperature "waves" with more or less sinusoidal tempera-
ture changes with peak-to-trough amplitude of about 1°C and wavelength of the
order of two long city blocks were observed with weak inversions (.5C/1.5 m)
or with zero lapse rates.
Summary of Results
The overall strength of the Fairbanks heat island appears to be strongly
tied to the strength of the low-level inversion in the undisturbed surround-
ings. The large values found for the maximum heat island intensity of this
59
-------
relatively small city are due to the well known fact that Fairbanks inver-
sions are among the strongest and most persistent in the world (Benson,
1970).
60
-------
SECTION 6
THE THREE-DIMENSIONAL HEAT ISLAND
MEASUREMENTS
Aircraft Observations
Four clear-night surface traverses were accompanied by helicopter flights,
and one, by an instrumented fixed wing aircraft. FAA regulations did not
allow us to fly at less than 300 feet ('v90m) over Fairbanks, and the two
flights under ice fog conditions were further hampered by poor visibility.
Instrumentation on the helicopter flights consisted of a thermistor
mounted on the side of the Pitot tube on the nose of a Bell 206 B Jet Ranger
helicopter. The thermistor output was recorded on a battery-operated
recorder. After the first flight, a small pressure transducer was hooked
into a second battery recorder, in order to provide precise documentation of
height changes. An experimenter in the back seat of the helicopter kept both
recorders running and noted location on the temperature chart. A second
experimenter rode up front as navigator. So long as the helicopter maintained
forward motion the thermistor was well ahead of the rotor wash, but when de-
scents were made to check lapse rates in outlying areas the speed sometimes
dropped to the point that the thermistor showed a sharp rise in temperature
as it was hit by the rotor wash. Such events were easy to identify on the
chart record.
The major problem encountered was in getting good altitude control.
Because of the very steep inversions and occasional step-like temperature
structure in the Fairbanks area (Holmgren et al, 1975), it was important to
know the exact height at which measurements were being carried out. On the
first flight, most of the runs across town ended 50 to 75m higher than they
61
-------
began, and only two runs had usable data. For this reason a pressure trans-
ducer was added to the instrumentation on the remaining flights. The trans-
ducer-recorder combination was calibrated against a high precision barometer;
pressure measurements were transformed to heights by using the hydrostatic
equation and observed temperatures.
The fixed-wing observations were made from a Cessena 180 with thermistor
attached to a wing strut. The flight coincided with Traverse 8, and ob-
servations were made at 320m, 470m, 780m and 1250m above the ground. Ob-
servations were attempted at 170m, but had to be discontinued because of
dangerously low horizontal visibility due to ice fog at that level. Figures
19 and 40 were taken during this flight.
Hi 11 -S1 ope Mea sur enient s
Previous work by the authors (Benson and Bowling, 1975) has shown that
2m temperatures on hill slopes are quite close to free-air temperatures at
the same elevation, although this is not always the case for measurements
taken on hill tops. As the initial stage in almost every traverse was a
descent from the road in front of the Institute to the main road in the flats,
60m lower, we automatically have the background inversion on most traverses.
Tethered Balloon Observations
Information on the background lapse rate is available from an earlier
project. Investigation of the surface inversions at Fairbanks using an
3
acoustic sounder included a number of ascents with a tethered 3.2 m balloon
at a fixed site. This balloon was shaped so that it ascended nearly verti-
cally and it carried apparatus for detailed profiling of the boundary layer
(Holmgren et al, 1975). The ascents were made at a location north of College
Road, marked A in Figure 2. Information on temperature, windspeed and pres -
sure were telemetered to the ground station, while wind direction was obtained
from the balloon orientation. Wind direction was not always available from
nighttime ascents.
-------
CO Concentrations
As CO is emitted primarily at street level, stable air should be re-
flected in a variation of CO levels with height. This was checked by
comparing 30 minute CO measurements at three heights with those measured at
the same site by the Fairbanks North Star Borough. The site is downtown,
less than a block east of the Second Avenue - Cushman Street heat core. The
Borough intake is located at 3m elevation, and instantaneous CO concentra-
tions were obtained using a Beckman 315 BL infrared absorption analyzer. The
instantaneous sampling was interrupted briefly to analyze our 30 minute bag
samples, collected at 10 cm, 1m, and roughly 12 m (building top height), so
all analyses were carried out using the same instrument.
RESULTS
The background lapse rate is known to be highly variable. Step-like
temperature profiles are common (Figure 34), with layers of constant tempera-
ture or even normal lapse rates being separated by thin layers with extremely
MARCH 6, 1972 ASCENT
'-A1 .'44 ?4S
"A
12345
Meters/sec.
Figure 34. Vertical profiles of temperature, acoustic sounder backscatfer
and wind velocity in a relatively undisturbed area (from Holmgren
et. al. 1975).
C3
-------
strong inversions. Strong ground inversions with isothermal conditions and
wind speeds of 6-9 m sec~ above 20 to 30 meters have also been observed, as
have isothermal conditions near the ground with strong capping inversions
several tens of meters aloft. On occasion a uniform inversion may reach to
100 m or more, as appeared to be the case during traverse 24.
Lapse rates in the lowest 90 m of the atmosphere in the city have had
to be inferred from ground temperatures and the consistent lack of variability
of CO concentrations with height. CO concentrations were identical at 10 cm,
1m, 3m and 12m in the city core under background inversion conditions, so
good mixing and an adiabatic lapse rate to 12m are indicated.
As pointed out in the methods section of this chapter, the temperature
in front of the Geophysical Institute, 60m above the elevation of the flats
on which the town is sited, is a good approximation to the free-air tempera-
ture at that height. Thus the difference between the temperature at the
Geophysical Institute and that at the cold background sites can be taken as
the mean background inversion over the first 60 m, TR (60) - TR(0). If the
heat island intensity, AT (0), for each traverse is compared with this value
for the same traverse, it is found that the equation
AT (0) = 1.6 + .75 (TB (60) - Tg (0)) ' (1)
predicts the heat island intensity within 1.5°C for all but 3 cases (Figure
35). Considering the complexity and variability of the structure of the
background inversion, the fit of the points must be considered excellent.
It is possible to use (1) to make some deductions about the height of
the crossover point of the Fairbanks heat island - that is, the height at
which the temperature is the same over the city as it is at the same height
over undisturbed terrain. We assume that the lapse rate over the city center
will not exceed the adiabatic rate. Then if the ground level city tempera-
ture is T (0), the city temperature at a height z will be no lower than
T (0) - .Olz, .01°C/m being the adiabatic lapse rate. If the city tempera-
\*
ture at 60 m is in fact given by T (60) = T (0) - .Olz, it can readily be
\s \j
shown that the heat island at 60 m, AT(60), is given by
AT(60) - 1 - .25 (TB (60) - TB (0)) (2)
For AT(60) -0, i.e., for (Tg (60) - TB (0)) <_ 4°C, the assumption of an
64
-------
8 10 12 14 °C
TB (60) - TB (0)
Figure 35. Comparison of heat island intensity and approximate 60 m inversion
strength (Geophysical Institute temperature minus the same back-
ground temperature used in computing the heat island intensity).
adiabatic lapse rate over the city is not unreasonable and we can safely
say that for a 60-meter inversion of 4°C or less, the heat island normally
extends to above 60m.
For a 60-meter inversion of more than 4°C, (2) predicts a negative heat
island at 60 in. This would represent convective overshoot and, while it may
occur over a limited height range, Eq. (2) should not be considered to have
any precise physical meaning for (Tg (60) - Tg (0)) > 4°C. If, however, we
assume that the background lapse rate, y, is constant at least to 60m and
given by y = -(Tg (60) - Tg (0))/60, we can define a crossover height. Let
AT (z) be the heat island intensity at height z. We have the city temperature
at z, T (z) = T (0) - .Olz, and the background temperature at z, Tg (z) =
TR (0) - Yz. So
AT(z) = (Tc (0) - .Olz) - (Tb (0) - Yz).
(3)
G5
-------
But Tc (0) - TB (0) = AT (0) = 1.6 + .75 (Tg (60) - Tg (0))
and (Tg (60) - TB (0)) = -60y from the definition of y. so
uT(z) =1.6 - 45 Y - (-01 - Y)z. (4)
The crossover is defined by AT(Z) = 0, so the crossover height Z, is
given by
Z = (1.6 - 45Y) / (.01 - Y) (5)
Representative values of Z for several values of Y are summarized in Table 6
and Figure 36.
TABLE 6. DEDUCED CROSSOVER HEIGHT,
Z, AS A FUNCTION OF UNIFORM BACKGROUND
INVERSION STRENGTH IN °C/100 m.
Y, °C/100m
0
-1
-2
-5
-10
-15
-20
-30
Z, m
160
102
83
64
55
52
50
49
If the values of z in Table 6 are compared with those calculated for
a constant lapse rate in Section 9, the heights deduced during the stronger
inversions are found to be too high relative to the fairly well substantiated
value of z = 60m for (TB (60) - TB (0)) equal to 4°C. These stronger
inversions are already known not to be constant with height, so the lack
of agreement is no surprise. No detailed calculations have been made for
more realistic lapse rates, as possible forms are too numerous. However, a
concave-upward shape (steepest inversion near the ground) gives the best
qualitative agreement between the theoretical and observed data. Figure 37
illustrates both the basis for the calculation and the effect of a non-
constant background lapse rate.
66
-------
150r
-40
-30 -20
LAPSE RATE, X, "C/100M
01
Figure 36.
Plot of crossover height against inversion strength based on
empirical relationship between heat island intensity and 60 m
inversion and assumed constant background lapse rate.
Assymptotic approach to Z - 45 m for strong inversions is
physically unrealistic and indicates probable non-constant back-
ground lapse rates for these conditions.
The discussion to this point has assumed an adiabatic lapse rate above
the city core. Although a superadiabatic lapse rate is unlikely on theo-
retical grounds, a stable lapse rate is quite possible, especially if any
substantial fraction of energy transfer to the air is due to elevated sources
(e.g., building walls, windows, elevated stacks). Thus the calculations to
this point (allowing for modification under strong inversion conditions, see
Figure 37) give a lower limit for the vertical extent of the heat island. An
upper limit must come from the aerial traverse data. The aircraft measure-
ments in every case showed that, at the lowest level sampled over the city,
temperatures were above those at the ground, so that inversions, probably
67
-------
with bases above ground level, were present. In addition, city temperatures
at levels accessible by aircraft were not measurably above those over the
background areas, e.g., AT (z) = 0 when z is the lowest flight level: 90 m
under clear-sky conditions, and 170m in the one dense-fog case sampled. For
the clear-sky case, then, the lapse rate within the city can be bracketed by
an adiabatic lapse rate to around 50m with an inversion above that level as
one extreme case, and by a weak inversion from the ground to at least 90m as
the other extreme (Fig. 37). A near-adiabatic lapse rate to around 50m with
an inversion at higher levels seems the most likely solution over the heat
island core.
100r
UJ
x
20-
TEMPERATURE - TB (0), °C
Figure 37. Limiting values for city lapse rate. Heavy dots are measured
temperatures; light line is the adiabatic lapse rate over the city;
dashed line is the background temperature profile assuming a
constant lapse rate to 60m; dash-dot line is a more realistic back-
ground temperature profile. Cross shows calculated crossover
height, circled cross,a more realistic value. The true city
temperature profile must lie within the triangle formed by the
heavy solid lines.
Measurements in clear weather above 90m showed an apparent influence of
the city to and above that level, even though there was a 1°C to 6°C in-
version between 2\\\ and 90m over the city center. Both turbulence and wave-
60
-------
like temperature fluctuations differed in and out of town. Turbulence was
encountered only over the city area, and was frequently associated with
temperature fluctuations (breaking gravity waves) in situations when no such
fluctuations were encountered over background areas. Flights during which
temperature waves (always without detectable turbulence) were encountered in
the background areas generally had relatively steady temperatures over the
city. It appears in these cases that relatively low-level convection in the
city may be acting either to influence the amplitude or stability of pre-
existing waves or to generate waves where they did not previously exist. In
the latter case (Figure 38) the wavelength in the west part of the east-west
wave train appears similar to the distance between intersections on Airport
Road, while the eastern section was over a loop of the Chena River. In each
case, stronger than normal heating near the ground could have excited deeper
than normal convective mixing which in turn could generate upper waves in
somewhat the same way as a cumulus tower may form a pileus or scarf cloud by
coherent lifting of stable air.
Figure 38.
Thermistor traces across town during Traverse 23 at an elevation
of 90m. Left side - North to South and return along Cushman
Street; right side - East to West along the line of Airport Road.
The lowest aerial traverse height used in constructing cross sections of
temperature above the city during traverses 24 and 29 showed sharp temperature
gradients across the city. The colder ground areas north of the city had
substantially higher temperatures aloft than did the background areas to the
-------
south, which were warmer near the ground. This was at first also interpreted
as being due to coherent lifting of upper, stable air layers by convective
activity lower down in the city air. Attempts to consider the amount of
lifting which would be necessary in specific cases, such as Traverse 24
(Figure 39), together with the realization that while ground level winds were
generally from the north, winds at flight level were more often from the
east, made this hypothesis untenable in its original form. As both cases
with definite gradients aloft showed gradients in surface background temper-
ature of opposite sign, we suggest that the difference is due to more effec-
tive turbulent mixing of heat down to the ground in the southern area, which
is relatively exposed to east winds (note topography on Figure 2). If this
is the case, the position of the sharp gradient over the city is fortuitous.
The horizontal visibility above the city was significantly impeded
during both aerial traverses with ice fog. During traverse 29, with moderate
ice fog, the helicopter was in ice fog at an elevation of 90m, although the
visibility was not so low as to terminate the measurements. An attempt to
-4.5
-4.0 -3.5
-3.0
0
Imite
Figure 39. North-South temperature cross section of the air over Fairbanks,
based on data obtained midnight to 0100 March 14, 1975 (traverse
24).
70
-------
measure temperatures at 170 m during dense ice fog, however, had to be
terminated due to dangerously low visibility (traverse 8). Large-scale
vertical mixing to the flight levels seems unlikely, as the measured tempera-
tures aloft were still higher, if only by 1°C in 29 and 2°C in 8, than any
observed at the ground. The spread of major plumes (e.g., those from power
plants) is known to produce fog layers aloft which at times blend with the
ground-based fog over the southwestern (downwind) part of town, and the
aircraft were probably in plume-derived fog. However, there is other evidence
for an increase in the depth of vertical mixing during ice fog. SCk* which
is emitted primarily from elevated stacks, shows tremendously enhanced
ground-level concentrations under ice fog conditions (Holty, 1973); while CO,
which is primarily from ground-level sources, decreases at ground level under
the same conditions. The combination of SO- and CO trends strongly indicates
increased mixing depths during ice fog, as predicted by Bowling (1970). The
major difference from this earlier prediction is that mixing is now known to
take place over the city even when skies are clear.
In summary, a normal lapse rate or very weak inversion probably extends
to at least 50m within the city, the height increasing as the background
inversion weakens and/or ice fog develops. Temperature disturbances at
higher elevations exist, but are not connected simply with the low-elevation
heat island. With clear skies and no ice fog, the primary effect of the heat
island on the air above 90m height is on air motion, which in turn may lead
to temperature differences between the air over the city and its surroundings.
71
-------
SECTION 7
THE WIND FIELD
MEASUREMENTS
Fi xed Anemometers
Two anemometers are routinely operated in the area, one at the National
Weather Service (WS in Fig. 2) and one operated by the Fairbanks North Star
Borough on the south bank of the Chena River (B in Fig. 2). Observations
were available at hourly intervals from the Weather Service, but their
instrument normally did not record winds of less than 1 m sec" . The data
from the Borough instrument were available only as output from a computer
program which produced "average" hourly winds. During the second year of
the study we added three more instruments. These consisted of a Lamprecht
wind-run recorder installed at Peger Road (PR on Fig. 2), and two Gill
propeller-vane anemometers, one at Creamers Field (CF on Fig. 2) and one at
North Slope Batteries, (N on Fig. 2).
Aenionieters
During traverses 17 and 20, wind speeds were measured throughout the
area by six to ten volunteers with small, hand-held anemometers. The
anemometers, which had starting speeds of a few cm sec" , were fabricated in
the Geophysical Institute Shop. Records from an earlier series of measurements
carried out by Benson and Weller in the late 60's were also available. The
primary difficulty with this technique was in scheduling. Traverse 17 was
run under an overcast sky and 20 was run under a stratocumulus layer which
72
-------
was breaking up and re-forming in a thoroughly erratic manner. The earlier
observations were made under light ice fog conditions in 1969-70.
These studies demonstrated that the horizontal scale of some major wind
shears was less than the spacing we could hope to obtain with any reasonable
number of point observations. This, of course, affected the fixed anemometers
as well. The information obtained was worthwhile in demonstrating this
variability, as well as providing some data on very low wind speeds, es-
pecially in the northern part of the area.
Pljjme £rvft_ Directions
Power plant stack plumes are generally visible from observation sites
above the ice fog, and have previously been used to deduce wind directions
above fog top height. Some caution is necessary in interpreting this type of
data as differences in wind directions indicated by two or more plumes which
do not reach quite the same height may be due to vertical shear. This must
be considered even when only high stack plumes are used (Figure 40).
At temperatures between -20°C and -30°C moist plumes from heating plants
in single-story buildings and from vehicles were generally visible. Low-
level wind directions inferred from the direction of these plumes were
recorded as opportunity offered on traverses or while experimenters were
driving around in the area for other purposes. Although the effects of time
and space could not be clearly separated in observations of this sort,
certain patterns did show repeatedly enough to be useful.
A broader scale view is possible using time-lapse photography of plumes.
This technique has been used experimentally in the Fairbanks area for more
than ten years. The power plant plumes are clearly visible in area-wide
shots, but resolution of low-level house plumes required a telephoto lens and
a rather limited field of view. Vehicle plumes were not generally visible.
Plume observations, whether direct or via time-lapse photography, are
possible only over a very limited range of conditions. Small plumes rarely
have enough condensation to be visible at temperatures above -20°C, and
below -30°C the plumes tend to be lost in a thickening layer of ice fog.
Nevertheless, plume observations are one of the best currently available
methods of observing the wind field, as opposed to the wind at individual
points.
73
-------
Figure 40. Photo of Fairbanks area, looking east, taken 3 January 1975 with
a temperature of -42.5° measured at the city core. East, west,
and north winds roay be observed to affect plumes emitted at
different levels.
-------
RESULTS
Poijrt Measurements
Winds measured in Fairbanks tend to be either several m sec" and fairly
consistent in direction or under 2 m sec" and extremely erratic. The first
case is associated with relatively good dispersal conditions and slight heat
islands and is of no particular interest with regard to this study. The
second case can be illustrated by a day's records from our three anemometer
sites (Figure 41). Figure 42 illustrates for comparison a few possible sets
of wind directions which give coherent areal flow patterns, or which persist
for an hour or more on the wind record. There were also two periods in which
relatively uniform northeasterlies prevailed. As the spacing between stations
was 5 to 7 km, air moving at the observed speeds would have taken 2 hours or
more to travel between stations. Apparent patterns which last for less than
2 hours are probably due to local eddies or waves (this includes most of the
available observations).
The particular day shown in Fig. 41 was selected because the combination
of low wind speeds and sufficient constancy of wind direction at individual
sites allowed the 15 minute averages to be defined. Many days with low wind
speeds had winds so unstable in direction that no readable plot could be
made.
Simultaneous measurements at several points using hand-held anemometers
provided additional data on wind fluctuations, as well as some suggestions of
wind fields, although not under the best sky conditions. The mean wind field
from Traverse 17 was shown in Figure 25; Figure 43 shows the wind fields from
two different measuring intervals during Traverse 20. The two stations near
College Road in Figure 43 are of particular interst, as they maintained
virtually a 180° shear for the entire 40 minutes of measurement. The record
from the southeastern station (Joy School) includes a notation that when wind
speed dropped, the direction would change temporarily from WSW to NNW.
Evidently some sort of kilometer-scale shear or local eddy was maintaining
itself with relatively slight shifts in position for almost an hour in this
area.
75
-------
WIND FIELD Feb 28, 1976
Creamers Field
Peger Rd.
North Slope Battery
Weather Service
®
0)
to
|W
:\ ,
w
M I 2 3 45 67 8 9 10 II N 13 14 15 16 17 18 19 20 21 22 23 M
Alaska Standard Time, 28 Feb 1976
Figure 41. (above) Comparison of winds at three sites through the course of
a day.
Figure 42. (below) Suites of wind directions for various simple wind fields
for comparison with Figure 41. Wind measuring sites coded same
as Figure 41.
S
E
N
W
S
0
—
o
X \
—
t
/ \
/• N
—
14-16
J,
11-13
/
\ t
-------
Figure 43. Wind vectors for 2 two-minute measuring periods during Traverse
20 (0900-1000 8 February 1975). The lengths of solid arrows
indicate measured wind speeds; dashed arrows indicate that only
wind direction was available. Filled arrowheads-0925; open
arrowheads 1000.
Figure 44 shows near-surface (2m) wind roses representing wind directions
and strengths measured during four periods of light ice fog, when the
stratification of the atmosphere is stable and the surface air flow pronounced.
All four ice fog events occurred during the winter of 1969-70. As can be
seen, the wind speeds and therefore the diffusion processes close to the
ground are low. The flow of air is determined by the local topography and
results from gravitational downslope (katabatic) drainage of cold air, produced
by radiative cooling of the surface. River and stream valleys, even if
small, thus clearly become channels draining air downslope, as can be seen
from Figure 44. The slope angles of either valley or hi 11 si ope determine the
77
-------
NO OF "'« FREQUENCY /
OBSERVATIONS-/^
0-90 90-WO 100-BO 150-200 >200
1 •••
WIND SPEED (CM/SEC)
12 KM
Figure 44. Low level wind roses at sites in the Fairbanks area for 4 periods
of light ice fog in 1969-1970.
wind speed: strong winds are generated on steep slopes; weak winds on re-
latively flat terrain. Frictional effects resulting from high surface
roughness probably also affect the wind speeds.
Winds are generally less than 0.5 m sec" (1 mph), but occasionally on
steeper slopes or in well-defined stream channels they reach or exceed 1 m
sec" (2 mph).
The mean wind speed near the surface in Fairbanks and surroundings
during periods of ice fog is approximately 0.5 m sec" (1 mph). If this wind
speed were uniform, an air parcel near the surface could be expected to
travel about 40 km day~ down the Tanana valley. However, the problem is
more complex because of the oscillation of low level air over the city which
reduces the net value of 40 km day" to a significantly lower value.
73
-------
Vertica1 and Small-Scale Horizontal Shears
Both the acoustic sounder and plume observations have confirmed the
existance of substantial wind shear in the vertical. Figures 34 and 40
document the types of vertical wind structures likely to occur. In addition,
we have observed cases in which plumes from idling vehicles and those from
nearby single-story buildings have indicated substantial (180°) wind shear
between about 1 m and 4 m.
Horizontal shears are indicated by 180° wind shifts indicated by plume
directions over distances of 100m along roads. Small-scale convergences are
sometimes indicated by vehicle plumes—in one case, plumes from cars idling
in parking lots on opposite side of College Road converged on the road.
Time lapse movies and visual observations of house plumes have documented
both organized and apparently chaotic variability in wind direction in space
and time. In one striking case, a 180° change in wind direction was observed
to propagate back and forth along a row of houses several times. Presumably
n gravity wave was responsible.
It should be emphasized that observation of wind shears through plume
observations requires a good deal of luck--plumes must be present on both
sides of a shear zone, the temperature and humidity conditions must be such
that the plumes are visible, and an observer must be present who recognizes
the significance of the plume directions.
There is other evidence that periodic motions exist regularly and
frequently in the lower atmosphere over Fairbanks. Benson (1970) observed
and photographed waves at the upper boundary of the ice fog layer and
speculated on the wind shear and diffusion processes at this boundary.
Wilson and Fahl (1969) defined the characteristics of the waves and identified
them as atmospheric gravity waves. Using infrasonic pressure transducers,
they found the mean periodicity to be approximately three minutes and the
mean pressure, temperature and orbital wind speed amplitudes were 15 microbars,
0.5°C and 0.2 cm sec" , respectively.
A long period motion of air has also been observed by measuring
temperature oscillations on the slopes of hills which surround Fairbanks on
three sides. These temperature variations have periods of roughly 20 minutes.
Benson (1970) and Haurwitz (1973) have interpreted them as internal seiches
79
-------
in the pool of cold air confined by the local topographic features.
The Low-Level Horizontal Wind Field
It is possible to organize the data we have gathered according to the
following working hypotheses:
We consider that the low-level wind field at a particular time is due to
the interaction of a number of driving mechanisms, each of which tends to pro-
duce its own pattern of winds. Observations generally show a combination of
two or more of the patterns discussed. The eight most important patterns are:
1. Uniform wind direction at relatively high speeds.
2. Strong winds south of town, stagnation further north.
3. Northward penetration of gravity drainage down the Tanana River.
4. Gravity drainage from the local hills.
5. A back eddy from the Tanana drainage.
6. Covergence toward the heat island.
7. Gravity waves.
8. Local wind shears.
(1) Uniform wind direction at relatively high speeds. Pattern 1 is
associated with weak-inversion or neutral lapse rates and probably
represents low-level penetration of geostrophic winds. It gives
good dispersion conditions and little heat island. The geostrophic
wind probably has a major influence on the relative importance of
the other wind fields, even when penetration to the ground does not
occur.
(2) Strong winds south of town, stagnation further north. A modified
version of pattern 1, which we will call pattern 2, has high wind
speeds in the southern part of the Fairbanks area with relatively
stagnant air in the more sheltered areas farther north. Pattern 2
can cause a major breakdown of air quality forecasts, as the Weather
Service Office at the airport is frequently in the southern, high-
wind zone while the major part of Fairbanks is in the northern,
stagnant zone. The windy area is normally much warmer than the
stagnant area, especially if skies are clear. This results in a
"one-sided" heat island, such as was observed in Traverse 38
(Figure 24).
80
-------
(3) Northward penetration of the gravity drainage down the Tanana
River. This may almost be considered a special case of pattern 2,
as it is unlikely to occur without reinforcement from easterly
geostrophic winds aloft. So far as the heat island and dispersion
conditions are concerned, the effects are similar to (2), but may
be weaker.
(4) Gravity drainage from the local hills. In broad scale, the
combination of topography and Coriolis acceleration tends to give
very weak east to southeast winds in the northeast part of the
Fairbanks area (along the Steese Highway), weak northerly flow near
the University, and weak northeasterly flow southwest of town.
Observations of wind directions along College Road in near-calm
conditions generally confirm this mechanism (but see also patterns
5 and 8). Observations along Airport Road are less consistent,
probably due to more interference from other patterns. On a finer
scale, hillside areas with different slopes, aspects, and elevations
will tend to produce air of different temperatures which would be
expected to move out over the basin at different heights. Observed
"low speed jets" moving in different directions at different heights
(Holmgren et al., 1975) could be due to this mechanism.
(5) Back Eddy from the Tanana River drainage. This pattern, with a
clockwise eddy, was observed by Benson and Weller in the late
1960's and described by them in an unpublished report to ARCO and
to Earth Resources Company in 1970. Their data were given in Figure
44, and their summary figure of the wind structure is included here
as Figure 45. The north to south component of flow in this pattern
would act to reinforce the same component in (4), while the sense
of rotation in the two cases is opposed. Combination of the two
patterns could give eddies in either or both directions superimposed
on a general drainage flow from the north.
(6) Convergence toward the heat island. Southerly components were
observed on several occasions along Airport Road while air flow on
the College Road section was northerly. Also, the simultaneous
wind measurements taken on January 25 (under nearly calm but
cloudy conditions) show convergence on the city center from all
81
-------
Figure 45. Surface flow pattern for ice fog deduced from the data in Figure
44.
directions. The interaction between this pattern and (3) is of
particular interest with regard to the proposed industrialization
of the North Pole area, which is located in the Tanana air drainage
upstream of Fairbanks. Although the undisturbed air drainage down
the Tanana Valley would normally carry pollutants well south of
Fairbanks, convergent flow toward the city from the south could
carry pollutants back northward into the city core.
The six patterns discussed so far are reasonably stationary in time and
have a broad spatial scale. Unfortunately for purposes of measurement, they
are complicated by the presence of at least two types of small-scale and/or
time-dependent wind fields with local speeds comparable to (2) through (6)
above (0.2 to 2.0 m sec~ ). These are listed as (7) and (8) below.
(7) Gravity waves. Waves with periods of 2-3 minutes and wavelengths
of several hundred meters are well known from infrasonic and
acoustic sounder records (Fahl, 1969; Holmgren et al., 1975) and
are probably responsible for the usual "ocean wave" appearance of
the upper boundary of the ice fog layer as seen from hilltop
-------
locations (Benson, 1970). Several 180° wind shifts in succession
were at times observed along Airport Road with wavelengths com-
patible with a gravity wave orign. Time lapse photography of
several house plumes south of the University also suggested wind
fluctuations of gravity wave orign, and some rhythmic fluctuations
of temperature along traverses and on thermograph records (Benson,
1970) may be due to the same mechanism. Longer period gravity
waves are believed to be associated with seiche oscillations between
Chena Ridge and Birch Hill ( ^ 30 minute period) and between Chena
Ridge and the Salcha Bluffs area ( ^ 4 hour period) (Haurwitz,
1973). These long-period standing waves do not seriously interfere
with wind measurements, although they could be of considerable
importance in pollutant transport. This is especially true of the
4-hour wave with respect to effluents from the North Pole area.
The high-frequency traveling waves, however, introduce considerable
confusion into both wind and temperature records. All of the
gravity wave phenomena are probably dependent primarily on the
geostrophic winds aloft for excitation.
(8) Local wind shears. This category is rather poorly understood. Its
existence is based primarily on an observation of a 180° horizontal
wind shear over a distance of less than a kilometer and remaining
stationary for at least an hour (Figure 43). Certain temperature
measurements (e.g., persistent cold and warm spots not obviously
controlled by topography and not constant in position from day to
day) are also best explained by some sort of channeled air drainage.
Most of these observations (including the measured wind shear) were
made along College Road, which frequently is nearly paralleled by
an isothermal line in a zone with a strong horizontal temperature
gradient. The observations may indicate: (a) kilometer-scale
eddying; (b) wind shear across a micro-scale cold front upwind of
the heat island, possibly with miniature wave cyclones; (c) shear
between regions controlled by different wind patterns (1) through
(6); (d) a very low level vertical wind shear on a slightly sloping
interface. Other explanations are undoubtedly possible. The
phenomenon is of interest with respect to mechanics of air flow
83
-------
into the heat island from the north and because of possible inter-
ference with measurements of the regional wind field. Wind ob-
servations along a traverse cannot distinguish between (7) and
(8), so the relative importance of these two patterns is unknown.
Radiation conditions, regional winds aloft, and surface pressure gradi-
ents probably combine to control the relative importance of the different
factors.
In addition to the horizontal patterns described, vertical shears are
extreme. In one photographically documented case, (Figure 40) the upper
stack plumes were under the influence of a strong east wind, lower level
plumes from the same power plant were being driven by a west wind, and plumes
from 2-story apartment buildings and a school were going south. All three
plumes were withjji an exceptionally deep (200 m) ice fog layer.
In summary, air movement in the Fairbanks basin is controlled by a
number of competing driving mechanisms. As a result, the observed wind field
is highly variable in both horizontal and vertical space and in time.
84
-------
SECTION 8
FAIRBANKS ENERGY-USE INVENTORY
The energy-use inventory concentrated on four energy sources: coal,
gasoline, fuel oil, and imported electricity. Locally generated electricity
(plus the waste heat released in the course of generation) was included in
the coal and fuel oil figures. The area covered is approximately that of the
Fairbanks North Star Borough; roughly 75% of the Borough population lives in
Fairbanks and most of the remaining 25% work and/or shop there.
ELECTRICITY
Electricity is imported 164 km from the mine-mouth plant at Healy by the
Golden Valley Electric Association (GVEA). GVEA figures for electricity
produced at Healy since the plant opened in February 1968 are given in
Table 7 and may be seen in terms of the area's commercial electrical gener-
ation in Figure 46. Like most in this section,this figure is cumulative,
i.e., the top line gives the total GVEA and Municipal Utilities generation,
the next line down gives total GVEA production, etc.
COAL
Coal is imported from Healy via the Alaska Railroad (ARR), which provided
us with monthly figures from mid-1967 to the present. In addition, monthly
consumption figures since 1964 were obtained from all of the coal-fired power
plants in the area--GVEA, Municipal Utilities System (MUS), the University
heating plant (U of A), and the two plants on Fort Wainwright. These are
shown in Figure 47. Comparison of the railroad and power plant totals on a
8'j
-------
~T~
i^
Ll_
0
00
—r
-^.
O
1— (
_l
_J
1 — 1
s:
z:
i — i
oo
i^
z:
<:
CO
Qi
-
1—
(-H
c_>
1 — 1
o;
i
<_>
UJ
_i
UJ
[^
1 1 1
1
CO
<:
i—
LO
r*1**
cn
*d-
f^^
1^*-
cn
oo
ps^
l^1^
cn
CM
r-^
cn
r-^
cn
CD
cn
cn
if")
cn
co
1 1~\
**LJ
cn
to co «3- i — LOLOi — r--LOUDr-~o
cou3cor^.i — LOCO^-I — r-^i-~-oo >=J-
oocM«y-oor-^i —
r^LOcn^^d-CMooooocMLOr^ CD
CDOO^CncMCM^-CO^i-CDi— O
r^-i — «d-LOOi — i — oocnLocDLO i —
CM^CnCMi — CDOOVJDCDi — CD
cOLOco«3-i— ocncMi— LOvD«d- oo
i — I — I — I— I — i — , — i — r— I — LO
oocDoocncMr^cocnoocMoo
UDCMOOi — OOLOCnOi — OOCMLO CD
«3-OOCMOOr-^LOLOCMOOUDf-CO
cocDcoLoi^cocor-xi-^cncM«^t- r^.
r- r— CTl
cxicocsjcMCDi — cn-^-CDi^^^t-
o>3-cocooor^r~-oo«3-coocQ oo
LO
(U >, . 4-> • • • S-
c-QS-i->,c:i— cr)Q.-i->>o ra
rt3CUroD-rt3r333a)OOCU OJ
•t)Li_2:ca:5:'-D'-Dca:oooz:Q >-
86
-------
50
45
4O
35
30
25
15
IO
0
• Old Fairbanks plant
Q Healy plant
£2 Fairbanks peaking complex
n Municipal Utilities
GVEA
1956 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75
Figure 46. Commerical electricity production as a function of time in and
for the Fairbanks area.
87
-------
CO
c
o
CO
•o
re
Q}
rO
CO
J^
c
(O
-Q
IO
U-
cu
c
ns
O.
l.
O)
O
CL
c
o
O
o
to
o
o
CJ
U S3 [ ]
o o o
in •f rft
(oo-j muoiM «
cn
-------
monthly basis showed many months with more power plant use than railroad
imports. However, comparison of yearly totals allows some estimate of how
the residual, which includes home heating and probably some waste, has changed
over the course of the study.
The power plant coal consumption has remained quite constant. GVEA
ceased coal-fired generation in the Fairbanks area in 1971, shortly after the
Healy plant became fully operational, but MUS increased its capacity at about
the same time. Home heating has always accounted for a relatively small
fraction of the coal use, the greatest value being 72 x 10 tons (- 20% of
the total) during Benson's (1970) study. As Table 8 shows, the residential
coal use has since dropped to less than 1% of this amount. Table 9 shows the
monthly breakdown of use by the various power plants for the last two years,
and Figure 47 shows this breakdown back to 1957.
To avoid the effects of stockpiling on monthly data in the final sum of
monthly coal use given in Table 10 and used for the total energy use, the
power plant use for each month was multiplied by the ratio of the yearly ARR
imports to the yearly power plant consumption. This slightly overestimates
summer consumption.
GASOLINE
Gasoline consumption figures for the Fourth Judicial District (which
includes most of Alaska north of the Alaska Range) were obtained from the
Alaska Department of Revenue. These were broken down by highway, marine, and
aviation use, but not by area within the District. Luckily, a large percen-
tage of the population and most of the road mileage are in the Fairbanks
area. Alaska Railroad imports to Fairbanks were around 75% of the District's
taxable sales, and this figure was used to estimate the Fairbanks area gaso-
line use. (Tax figures were used rather than Alaska Railroad figures because
it was felt they would follow monthly fluctuations more accurately.)
Figure 48, which shows both the adjusted Fairbanks area use from the tax
data and the Alaska Railroad imports, shows the much greater month-to-month
consistancy of the tax data. Marine use (which is a small fraction of the
total) was included with highway gasoline. Aviation gasoline is consumed
over a much wider area than is highway gasoline. Since most bush flights in
39
-------
TABLE 8. FAIRBANKS AREA COAL USAGE BY YEAR IN THOUSANDS OF TONS
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
* Thi<; fi
Alaska Railroad
Imports
292.6
283.4
322.0
350.7
334.5
Records
Destroyed
384.7
366.5
349.7
364.3
335.2
331.6
347.7
393.1
inurp i <; nhvinuO v no"
Power Plant
Use
243.2
281.6
256.8
278.2
284.5
302.6
316.1
348.6
350.8
338.5
335.0
335.1
354.3
331.9
331.0
347.0
378.9
h thp r.nrrprt. nnp
Residual
(Home Heating)
49.4
1.8*
65.2
72.5
50.0
46.2
31.5
14.6
10.0
3.3
0.6
0.7
14.2
for rpsidpntial
use, and probably represents a sharp decrease in stockpiling
in the area.
90
-------
TABLE 9. MONTHLY POWER PLANT COAL USE BREAKDOWN
IN 103 TONS - 1974 AND 1975
Ft.
MUS %
1974
1974
Year
1975
1975
Year
J
F
M
A
M
J
J
A
S
0
N
D
I
J
F
M
A
M
J
J
A
S
0
N
D
:
12
10
11
9
7
7
6
6
9
10
13
14
121
13
11
11
10
9
9
8
8
9
n
13
15
132
.6
.1
.9
.5
.9
.4
.8
.8
.6
.8
.4
.8
.5
.1
.4
.9
.0
.5
.2
.9
.5
.1
.1
.7
.7
.3
32
28
34
38
39
45
37
36
42
34
37
32
35
29
29
32
35
36
41
46
42
34
38
35
35
35
UofA
4.7
4.3
3.6
2.2
1.6
1.2
1.1
1.1
1.4
3.0
3.7
4.3
33.2
4.4
3.7
3.5
2.6
1.8
1.5
1.4
1.4
3.4
4.1
4.0
4.2
36.0
% Wainwn'ght %
12
12
10
9
8
7
6
6
6
9
10
9
10
10
9
9
9
7
7
7
7
13
14
10
9
10
22
21
19
13
10
8
10
11
12
18
19
26
193
27
24
21
16
14
11
9
10
13
14
21
25
210
.1
.2
.6
.3
.7
.0
.5
.2
.1
.5
.5
.7
.3
.6
.7
.6
.0
.8
.6
.2
.4
.9
.3
.1
.3
.6
56
60
56
53
53
48
57
58
52
57
53
58
55
61
62
58
56
57
52
47
51
53
48
55
56
55
Total
39.
35.
35.
25.
20.
16.
18.
19.
23.
32.
36.
45.
347.
45.
39.
37.
28.
26.
22.
19.
20.
26.
29.
38.
45.
378.
3
6
1
0
2
6
4
1
1
3
5
8
0
1
9
0
7
2
3
5
3
4
6
8
1
9
91
-------
TABLE 10. TOTAL COAL USE IN THE FAIRBANKS AREA, UNITS 10 TONS
1967 1968 1969 1970
Jan. 46 47 49
Feb. 42 44 39
Mar. 37 33 37
Apr. 34 29 30
41
31
30
27
May 30 28 25 25
Jun. 23 19 24
Jul. 20 19 21
21
18
Aug. 22 23 23 20
Sep. 23 25 23
24
Oct. 33 32 28 31
Nov. 40 42 36 34
Dec. 43 45 35 46
Year- 394 ^8b 366 350
3 -
(1)
C
0
-L
I-
X 2 - - [
1 Jn'l
nfl ffl^i
%jyNi [
Ji n
jTjll 1
1971 1972 1973 1974 1975
54 40 39 39 47
37 37 32 36 42
37 37 33 35 38
27 27 26 25 30
25 23 21 20 27
25 17 17 17 23
20 17
21 18
17 18 20
19 19 21
24 24 25 23 27
28 27 30 32 31
34 32 34 37 40
39 37 37 46 47
364 335 33? 1/lfl W
Iflf
|
rjS If]
1
I
IV
«H
w
VJ
1967 1968 1969 1970 1971 1972 1973 1974 1975
Figure 48. Gasoline consumption in the Fairbanks area. Heavy line-adjusted
tax data; light line-Alaska Railroad imports
92
-------
the interior have Fairbanks as one terminus, 10% of the total aviation use
was assigned to the Fairbanks area. Both the raw data for the Fourth Judicial
District and the adjusted total for the Fairbanks area (0.75 x highway and
marine + 0.10 x aviation) are shown in Table 11. Note that while electricity
and coal consumption are overwhelmingly dominated by winter heating and
lighting requirements, gasoline consumption has a clear summer peak. A much
weaker and less consistent midwinter peak probably reflects the lower mileage
per gallon (due in part to increased idling) at temperatures below about
-25°C.
Gasoline data from periods before 1967 had been destroyed before we
requested it. However, Benson (1965, 1970) had earlier obtained values for
gasoline imports to Fairbanks from 1957 through 1964. His Figure 4 (which is
not cumulative) is here reproduced as Figure 49. Note that the full vertical
scale on this figure is equivalent to the first scale division on Figure 48.
ICE FOG. LOW TEMPERATURE AIR POLLUTION
900
800-
700-
600-
&
»<
£ 500
9
_J
3
400
2500
2000
O)
o
X
1500 I
o
3
*:
L .. .
ALASKA RAILROAD
F"
n
In
J U
\fU
kJ
JOO-
200
100
1000
XX)
, MILITARY PIPELINE
(hi
TRUCK SHIPMENTS TOFT WAINWRIGHT
I95t
1957
I9M
I960
IMI
1962
IW3
IM4
IM5
Gasoline imported to Fairbanks and Ft. Wainwrighl.
Figure 49. Gasoline imported to Fairbanks and Fort Wainwright prior to 1965
(not cumulative).
93
-------
FUEL OIL
Fuel oil is imported both by the Alaska Railroad (ARR) and by truck.
ARR data were obtained without difficulty, but figures for total truck
shipments were unobtainable. Yearly total figures from the Federal Energy
Administration for 1974, compared with ARR data, made it clear that truck
shipments could not be ignored. Total fuel oil consumption was therefore
calculated from data on use rather than imports.
Fuel oil is used as a motor fuel, as a heating fuel for buildings, and
for electricity generation (primarily as peaking power). Sources of data
were: the Alaska Dept. of Revenue for highway and marine diesel consumption
in the Fourth Judicial District; the eight fuel oil companies currently
operating in Fairbanks for heating oil; and GVEA, MUS, and Fort Wainwright
for power generation oil use.
Diesel fuel is used primarily for trucks and heavy equipment and has a
substantially greater role than gasoline in long-distance transport.
Furthermore, the use of diesel fuel outside of the Fairbanks area increased
greatly during both the initial transport of material to the North Slope in
1969-70 and the actual construction of the Alaska Pipeline. We estimated
that under normal conditions (e.g., 1967), 60% of the diesel fuel for motor
We also assumed that the ratio of diesel to gasoline use in the Fairbanks
area had a ceiling which was estimated from the greatest ratios of tax figures
observed in normal years. Table 12 shows these ratios year by year; the
ceiling ratio was set at 0.4. The years 1970, 1971, 1974, and 1975 were
adjusted to this ratio by multiplying the basic 60% figure by 0.4 divided by
the calculated ratio for that year. The multiplier figures for the fraction
of the total diesel use in the Fourth Judicial District assumed to be consumed
in Fairbanks are also given in Table 12.
use would be consumed in the Fairbanks area (approximately a 50 km radium).
94
-------
3
o
g
I
Adjust
Total
Marine
Highwa
Marine
Highwa
.
II
CM o o ro **• r*. CM CM in r%. r-* in
On^DCO^-r— O^a^-COl/lf— CO
Or-.' — CM*J-**-r>.tor-.eo^^-
onooco^fr-*'— cncnr*. on
co in .— o ro CM oo *o in e\i
NOOOcorOf— CMr-.ooon cvj
,— i — .— CM CM CM •— i — , — CM
.
,_,__, — , — CM <«• ro ro 10 *— f— ^- o
.. *
r— r— o m *n ro .— in o* ir> r-. to
O cvi f* CM \O V** iO \O CM O r— •—
SCM r- cr» CM CM op r- •— *t 01 op
t£)ir)tcoj^-cnr-ooc\ioovo
^- r- CO ro PO ^- CM CM CM r—
r- ro r— o rt co r^- c
«£CftinkO^-Oi — CO »— r— r— • — \O
OOinvOOf— r- lOOT>Ch*O>— t**
^-, — , — r— CM ro ro CM i — (— o
i O Z Q >-
m ro co m ro o on .— o oo o> vo f—
on^oincMonontncv(O«ff— i — en
CMCM*— ro^-t0coOoOinin*4- ro
O r*» CM roror^07*O^^-ri««*1
r^.u>m^<.onr— roxoroocnoo
^-i — i — . — r— CMCMrxJCMCMt— •—
r— oncM«— r— oovo«*^-^-
r« en oo oo on « in in ^o op
op op i
roror
rtcnoo-—
<•*• CM oo on r- o * in u> r- op «* CM
•«•«*•»— loininin^r^r— \o ** r^.
^oi\jw>inincn^-csjcMr-i/>r-. *—
.— i— »— •— i— .— CM CM CM t— .— ,— t—
r*. ^- in ro r*- CM ^ O r-* •— CM ro *»•
inioooinroroioi— f— cMr^^- in
»— f— ro co ro ro ^- O
Siomini
*O«— r—
^CMf— rocn in
^t^-tnmvo O
r--.iocnr-.Ot— •* CM CM ^- en on CM
.
oo CM en ro 3F oo *o \o in CM CM
f— omoorocn-^rr^mi— en
cncor— m«*^oc3t— iOLS)oor-.
.— »— • — CM ro ro CM CM .— CM
-
•*
en
tn r- on o on
o
cn
r— CM CM CM rO r
.*j-
en C
.
oo \o en CM ro
F r- ««r co oo ^ »— en en co r- ua on
j«— CM»— •— inooOr-»rocMCM o
r-CMroroincQCMCMinro*ocM on
-onr
«
§r-.oroopcnf-.fv.inrocMCM
cnooocnroin^-iOrO' — r—
s
r— i— J^ r- r- •— CM •— CM f— r- ,— On
i— coro'cnifCMCo«--CMcnor^ O
^*--hro^^^ **•
corocn^-C
^- »n r- O
to o z o >-
95
-------
TABLE 12. RATIO OF TAXED YEARLY SALES OF DIESEL OIL TO THOSE OF
GASOLINE FOR THE IV JUDICIAL DISTRICT. SEE TEXT FOR
EXPLANATION OF THE MULTIPLIER
1967
1968
1969
1970
1971
1972
1973
1974
1975
Fourth J. D.
Diesel/Gasoline
0.27
0.24
0.36
0.63
0.48
0.25
0.21
0.82
1.3
Multiplier
for Diesel
.60
.60
.60
.40
.50
.60
.60
.30
.20
Figures obtained from the Alyeska Pipeline Company confirm that their
fuel use out of Fairbanks from 1968-1975 was about 60% of the total fuel
taxed in the Fourth Judicial District for that period. However, it is not
known how this fuel was broken down between taxed fuel, fuel on which taxes
were refunded, and non-taxed fuel. (Much of the Alyeska use was off State
roads.) Nor are records available for Alyeska subcontractors. Taking all
this into consideration, the multipliers are probably about right, although '
the results could easily be off by 20 to 30%. This uncertainty is most
serious during the summer months, when motor fuel consumption peaks and heat
and power requirements are at a minimum. During the winter months, motor
fuel is usually a small fraction of the total fuel oil consumption.
Power plant consumption of fuel oil occurs primarily at the Fairbanks
peaking complex of GVEA, for which GVEA provided monthly data back to the
opening of the complex in 1964. MUS has some diesel generating capacity, and
figures were obtained from MUS beginning in 1967 for months in which oil use
exceeded 10,000 gallons. MUS use in other months was figured at 5,000 gal/mo,
96
-------
but this is small relative to the uncertainty in motor use. Fort Wainwright
also uses relatively small amounts of oil. The Fort Wainwright figures prior
to 1975 were actually for total diesel imports via the military pipeline.
Data for 1975 were actual power plant consumption figures. The two sets of
data agree quite well, although the pipeline data probably have poorer time
resolution.
Heating oil is the largest fraction of the total area-wide oil combustion
on a yearly basis and in all seasons except summer. Eight companies are
currently selling heating oil in the Fairbanks area. Three of these (including
the largest single dealer) supplied us with complete or nearly complete
monthly sales in gallons from 1967 through 1975. Of the remainder, two
supplied 1974 and 1975 monthly data, one supplied 1975 monthly data, one (the
smallest) supplied its 1975 yearly data, and one (apparently the second
smallest) was only able to provide data for August and September of 1976.
The 1975 monthly totals are therefore based primarily on real data and are
probably accurate to within 2% or about 20,000 gal/month. Monthly data for
the companies and months when real data were not available were reconstituted
by taking the ratio of available monthly sales figures to those of the largest
company for the same months and assuming the ratio held constant. Where
enough monthly data were available to show a difference in the form of the
annual curve, this was retained in the reconstruction. One company started
operations in 1972, and their figures for 1972 and 1973 were reconstituted by
assuming a linear increase in the fraction of sales from 0 in June 1972 to
the observed fraction in Janaury 1974. It must be recognized that the figures
for heating oil become steadily less reliable as they go back in time;
however, the three companies for which accurate data are available throughout
account for about 50'Z of the total.
Table 13 and Figure 50 show the monthly breakdown of fuel oil use since
1967.
Fuel oil data from prior to 1967 had been destroyed by some sources,
but again the Benson (1970) figures are available from 1957 through 1964.
Benson's Figure 5 is reproduced as Figure 51. Again, this is not a cumula-
tive figure. Note that the full vertical scale on this figure is equivalent
to the first scale division on Figure 50.
97
-------
TABLE 13. FAIRBANKS AREA FUEL OIL CONSUMPTION, UNITS OF 10
SEE TEXT FOR DATA SOURCES AND ADJUSTMENTS
GALS.
1967
1968
1969
1970
J
F
M
A
M
J
J
A
S
0
N
D
Y
J
F
M
A
M
J
J
A
S
0
N
D
Y
J
F
M
A
M
J
J
A
S
0
N
D
Y
J
F
M
A
M
J
J
A
S
0
N
D
Y
Taxed
(Motor)
Fuel
126
147
257
264
250
586
472
440
497
295
173
257
3765
167
197
323
124
325
846
486
406
351
280
177
124
3806
102
306
502
1053
1090
625
515
495
655
796
487
341
6965
269
429
835
1601
1069
1319
1290
1438
1687
1206
966
1380
13488
Adjusted
Motor
Fuel
75
88
154
158
150
352
283
264
298
177
104
154
2259
100
118
194
74
195
508
292
243
210
168
106
74
2283
61
184
301
632
654
375
309
297
393
478
292
204
4179
107
172
334
640
427
527
516
575
675
482
386
552
5395
Heating
Oil
1252
1125
792
607
415
258
215
209
506
766
936
1322
8406
1339
1169
809
685
379
384
258
306
618
827
1173
1514
9452
2076
1427
985
870
454
443
261
535
662
767
1426
1254
11160
1768
1473
1061
951
487
427
346
579
725
845
1224
1715
11601
Power
Military
101
105
97
46
34
21
13
13
13
46
42
34
563
50
21
71
0
4
8
8
13
8
21
13
8
227
8
17
4
13
8
0
8
8
63
8
21
8
168
8
17
4
17
4
4
4
4
4
4
4
147
223
Plants
MUS
53
26
5
5
5
5
5
5
39
54
49
179
430
280
79
23
12
11
5
5
5
5
18
53
138
634
176
98
44
16
5
5
5
42
35
57
98
169
750
216
104
78
64
42
33
30
46
42
5
5
17
682
GVEA
37
6
2
6
1
0
0
62
13
34
73
80
314
126
56
5
2
5
0
3
1
1
21
209
18
451
111
43
60
43
9
9
10
11
10
29
395
13
743
0
11
6
11
44
163
51
11
16
26
59
33
432
Total
1518
1350
1050
822
605
636
516
553
869
1077
1204
1769
11972
1895
1443
1102
773
594
905
566
568
842
1055
1554
1752
13047
2432
1769
1394
1574
1130
832
593
893
1163
1339
2232
1648
17000
2099
1777
1483
1683
1004
1154
947
1215
1462
1362
1678
2464
18333
ARR
314
304
847
781
1005
1066
1309
978
940
1228
694
582
262
572
907
1069
1877
1996
12414
1463
891
509
1611
1715
1079
637
1313
1445
1581
1787
1829
15860
2592
1361
2220
1959
1727
589
713
971
1155
1291
1230
1711
17519
(continued)
98
-------
1971
1972
1973
1974
1975
J
F
M
A
M
J
J
A
S
0
N
D
Y
J
r
M
A
M
J
,)
A
S
0
N
D
Y
J
F
M
A
M
J
J
A
S
0
N
1)
Y
J
F
M
A
M
J
J
A
S
0
N
0
Y
J
F
M
A
M
J
J
A
S
0
N
1)
Y
Taxed
(Motor)
Fuel
946
738
1574
673
251
628
2766
1338
507
396
185
231
10233
111
181
427
429
363
654
822
701
979
560
293
156
5645
374
139
100
110
455
602
700
738
606
544
254
264
4878
312
487
1758
808
1056
1752
2024
1867
2376
2339
2066
2823
19668
2279
2991
4250
4361
4008
4640
3959
4966
5780
5994
6267
3986
534?0
Table 13. (Continued)
Adjusted Power
Motor
Fuel
473
369
787
336
125
314
1383
669
253
198
93
116
5116
67
109
256
257
218
393
493
421
587
336
176
93
3387
224
83
60
66
273
360
420
443
364
326
153
159
2927
94
146
528
242
317
525
607
560
713
702
620
847
5900
456
598
850
872
802
928
792
993
1156
1199
1253
797
10684
Heating
Oil
2006
1797
1426
1071
769
276
237
489
655
811
1468
1946
12951
1910
1681
1473
1064
623
397
346
451
820
1064
1565
1671
13065
2591
1854
1485
1001
715
422
466
688
806
1372
1933
2278
15610
2549
2638
1908
1235
716
658
590
917
1004
1782
2106
2585
18688
3391
2572
2018
1769
927
925
803
992
1443
2105
3284
3322
23550
Military
8
4
8
4
8
4
3
4
4
3
5
6
62
6
3
4
3
6
3
8
4
4
4
4
8
58
8
8
8
8
8
8
13
8
13
8
13
8
113
8
17
8
8
21
13
13
13
21
38
42
34
235
54
19
22
17
0
0
0
0
24
24
36
41
237
Plants
MUS
5
11
15
5
5
5
5
29
85
5
5
5
180
31
5
82
15
5
5
5
66
20
16
5
5
260
33
5
5
5
5
5
104
133
13
5
14
98
425
25
29
5
5
5
11
13
236
47
36
5
5
422
104
46
12
66
8
4
2
5
4
23
24
89
385
GVEA
40
4
16
3
21
86
21
10
123
19
158
492
992
590
576
289
9
277
105
107
5
208
74
248
670
3158
1053
525
640
44
5
9
13
22
315
398
619
585
4227
681
770
236
145
26
4
25
67
197
170
620
1047
3986
1686
1008
1069
485
467
88
89
474
829
1159
2018
1781
11152
Total
2532
2185
2252
1419
928
685
1649
1201
1120
1036
1729
2565
19301
2604
2374
2104
1348
1129
903
959
947
1639
1494
1998
2447
19928
3909
2475
2198
1124
1006
804
1016
1294
1511
2109
2732
3128
23302
3357
3600
2685
1635
1085
1211
1248
1793
1982
2728
3393
4518
29231
5691
4243
3971
3209
2204
1945
1686
2464
3456
4510
6615
6030
46008
ARR
1846
1600
1356
1407
967
967
822
840
1495
2186
1583
1313
16382
1138
1125
1308
1591
1052
649
519
1344
1496
1410
1416
2393
15441
1400
1998
1221
702
793
1316
777
963
830
1885
1813
1414
15112
1710
1376
2155
742
1203
923
872
960
1063
1524
1902
2033
16463
3401
2495
2618
2730
2748
1432
1149
1735
1768
3863
3650
4229
31818
-------
67
Figure 50.
68 69 70 71 72 73 74
Cumulative fuel oil use in the Fairbanks area.
1
("Electric
jPower
(.Generation
^Heating
JQil
) Motor
[Fuel
TOTAL ENERGY
The total energy derived from these fuels was calculated in terms of
million KWH/month. Healy electricity was included as generated. Figures
obtained from Ft. Wainwright indicated that coal as delivered (including
dirt, non-coal matrix, snow, etc.) had an energy content of about 3500 KWH/ton,
which was used to convert coal consumption to total energy production.
(Note that this includes waste heat; actual electricity generation runs
closer to 1000 KWH/ton.) The conversion factors for oil and gasoline were
both based on heats of combustion of 10 cal/kg, which gave 34.8 KWH/gal for
gasoline and 38 KWH/gal for fuel oil. The resulting total energy use for
the Fairbanks area is given in Table 14, together with mean monthly tempera-
tures. The conversion factors given above were used on the data from Tables
7, 10, 11, and 13 to arrive at the data of Table 14 which applies to most of
the Fairbanks North Star Borough. The data from Table 14 are plotted in
Figure 52. Note that the monthly mean temperatures are plotted inverted to
facilitate comparison with fuel use.
100
-------
1000
900-
800
700
BOO
500
400
300
200
100
HITS
3000
2SOO
DC
O
o
2000
1500
woo
soo
~L.
MILITARY
PIPELINE
RAILROAD i TRUCK DATA
.1 '
J
Jl
I»M
1957
1958
1959
19(0
1962
1963
1964 1965
FIIC/ (ii/ imported fo (he Fairbanks area. All liwl tanks at Ladd Al'B were /"i//ed by the Air
Force before l/ic base wus turned over (o (he Army, resulting in abnormally hig/i monthly rates near the
end of /.%'«.
Figure 51. Fuel oil imported to the Fairbanks area prior to 1965 (not
cumulative).
BREAKDOWN OF ENERGY USE
Per capita energy use was calculated for five months—January, June,
and December of 1971 and June and November of 1975. The population of the
Fairbanks North Star Borough from 1969 through 1972 was 45,000 - 1,000, and
45,000 was used for the 1971 months. By 1975, the population was estimated
(by the Borough) at around 65,000. The resulting per capita energy con-
sumptions for summer and winter are shown in Table 15. Although there is an
apparent increase in per capita energy use from 1971 to 1975, the increase is
almost entirely in transport use. This is probably the least reliable portion
of the energy inventory and the portion most susceptible to systematic errors
associated with pipeline construction, so the increase shown should not be
taken too seriously.
101
-------
TABLE 14. TOTAL ENERGY USE IN THE FAIRBANKS AREA, MILLIONS OF KWH
MONTHLY MEAN TEMPERATURES ARE INCLUDED FOR COMPARISON
J
F
M
A
M
J
J
A
S
0
N
D
Y
J
F
M
A
M
J
J
A
S
0
N
D
Y
J
F
M
A
M
J
J
A
S
0
N
D
Y
Healy
0
0
0
0
0
0
0
0
0
0
0
0
0
0
4
7
6
3
8
7
4
4
6
1
8
58
6
6
6
6
8
7
6
7
8
10
13
15
97
Coal Gasoline
163
148
131
121
106
82
72
76
80
116
142
154
1390
166
152
114
102
96
65
64
82
88
111
145
155
1341
171
136
128
104
85
84
74
80
80
97
125
122
1283
1967
22
22
25
21
27
46
40
35
48
28
25
33
372
1968
26
23
32
25
36
40
49
43
43
40
33
34
425
1969
38
34
35
39
41
49
55
51
57
46
39
39
523
Oil
58
51
40
31
23
24
20
21
33
41
46
67
455
72
55
42
29
23
34
22
22
32
40
59
67
496
92
67
53
60
43
32
23
34
44
51
85
63
646
Total
243
221
196
173
156
152
132
132
161
185
213
254
2217
264
234
195
162
158
147
142
151
167
197
238
264
2320
307
243
222
209
177
172
158
172
189
204
262
239
2549
T°C
-26.2
-21.6
-12.3
- 0.2
7.6
16.6
15.5
14.6
8.2
- 4.0
-12.5
-18.6
- 2.8
-23.9
-20.6
-10.7
- 1.6
- 8.6
15.3
18.8
14.7
6.0
- 5.5
-16.5
-27.6
- 3.6
-32.6
-21.9
-12.1
2.3
9.6
18.2
15.2
9.9
9.5
1.1
-17.1
-15.6
- 2.8
(continued)
102
-------
Table 14 (Continued)
Healy Coal Gasoline Oil Total T°C
J
F
M
A
M
J
J
A
S
0
N
D
Y
J
F
M
A
M
J
J
A
S
0
N
D
' Y "
J
F
M
A
M
J
J
A
S
0
N
D
Y
16
13
13
11
9
5
10
11
12
13
12
15
140
16
15
17
14
11
0
9
12
12
15
16
15
153
17
11
14
16
10
11
11
14
9
16
16
15
61
143
110
106
95
89
72
65
69
85
109
119
160
1224
187
127
128
93
85
85
69
71
82
98
117
132
1275
141
128
128
94
80
59
60
64
83
94
111
130
1174
1970
41
35
42
45
44
58
60
56
58
45
37
41
563
1971
43
33
40
41
41
52
57
60
60
45
41
46
560
1972!
43
39
43
41
42
54
63
73
62
48
42
44
595
80
68
56
64
38
44
36
46
56
52
64
94
697
96
83
86
54
35
26
63
46
43
39
66
97
733
99
90
80
51
43
34
36
36
62
57
76
93
757
280
226
217
215
180
179
171
182
211
219
232
310
2624
342
258
271
202
172
163
198
189
197
197
240
290
2721
300
268
265
202
175
158
170
187
216
215
245
282
2687
-26.8
-13.3
- 6.2
0.0
11.0
14.4
16.9
13.8
4.9
- 7.3
-11.9
-23.2
- 2.2
-35.4
-20.3
-18.6
- 3.0
8.5
17.4
16.1
13.4
7.0
- 2.3
-17.5
-21.0
- 4.5
-26.9
-23.4
-19.3
- 6.2
8.5
15.2
18.0
14.9
4.5
- 2.9
-13.9
-19.0
- 4.1
(continued)
-------
Table 14 (continued)
Healy Coal Gasoline Oil Total
T°C
J
F
M
A
M
J
J
A
S
0
N
D
Y
J
F
M
A
M
J
J
A
S
0
N
D
Y
J
F
M
A
M
0
J
A
S
0
N
D
Y
17
15
10
15
14
12
13
14
10
13
17
17
166
17
16
16
12
15
14
14
15
11
18
18
17
184
19
17
18
17
12
16
16
15
12
18
18
18
194
137
113
117
92
75
61
58
67
86
105
120
129
1160
137
125
123
87
71
58
64
67
81
114
128
160
1217
163
145
135
104
95
82
71
74
96
107
140
164
1378
1973
38
26
41
43
49
49
61
57
60
47
49
52
574
1974
45
44
40
46
52
59
65
70
63
54
52
49
640
1975
51
54
70
80
82
104
113
114
113
112
no
90
1093
149
94
84
43
38
31
39
49
57
80
104
119
885
128
137
102
62
41
46
47
68
75
104
129
172
1111
216
161
151
122
84
74
64
94
131
171
251
229
1748
341
248
252
193
176
153
171
187
213
245
290
317
2786
327
322
281
207
179
177
190
220
230
290
327
398
3152
449
377
374
323
273
276
264
297
352
408
519
501
4413
-27.9
-18.6
-11.2
1.8
10.3
15.8
16.7
12.8
8.5
- 3.8
-18.1
-19.6
- 2.8
-27.0
-27.7
-13.6
1.6
10.8
14.8
17.5
15.1
10.8
- 5.9
-17.4
-24.1
- 3.8
-26.4
-19.6
-10.8
- 0.9
12.0
17.4
20.2
13.4
7.7
- 4.5
-22.2
-26.8
- 3.4
104
-------
January
, mean
-244
Coal
Imported Electricity
E3 Gasoline
D Fuel oil
1967 1968 1969 1970 1971 1972 1973 1974 1975
Figure 52. Total cumulative energy use in the Fairbanks area broken down by
fuel type and compared with mean monthly temperatures.
An additional breakdown of total energy use for 1973-75 is given in
Table 16, in which mobile sources (taxed motor fuel), dispersed stationary
sources (heating oil and electricity), concentrated stationary sources, and
the military base are listed separately. The first two categories are self-
explanatory. Concentrated sources were the MUS and GVEA power plants. (The
U. of A. plant is primarily a heating plant, and its entire output was in-
cluded in the dispersed power.) Comparison of generation and fuel consumption
figures for the major utilities gave approximately 9 KWH/gal of diesel oil
and 1000 KWH/ton of coal. The remaining energy is mostly dissipated locally
in stack gases (from oil-fired plants) and in cooling water released to the
Chena River (from coal-fired plants) although MUS does have some steam lines
to the downtown area. Using the previous total energy contents of the two
fuels, the waste heat for the two fuels separately was calculated at 29 KWH/
gal of oil and 2500 KWH/ ton of coal. The total waste heat, however, was
calculated as energy content of fuel minus the electrical energy generated.
lOb
-------
X — *
l_
<
LU
31 •
CO
1 — t —
CO Z
_J CO
Z CM
t— «
Q
Z
z :*:
z
LU CQ
3 a:
O •-'
Q- eC
LU
CD LU
o; i—
LU
^> ^y
CX* *— t
LO'
UJ
1
CQ
<£
(~
>-
i
3;
t—
z
O LLJ
S. ca
3C 0
0 —1
_J _J
0
O 31
cd cn
o
O LO
LU r-^
ai cn
,__
LU '
-r o;
t- LU
CQ
LL. z:
O LU
^»
UJ O
co z
^D
co
—1 CC £f
11 ill ry
Z CO - <3" »-H
ct: c2 x
=£ UJ CD
— i r> ry
z s: a.
\x
rO
ai
t.
ca
>,
i— cn
t3 t—
4-> QJ
o c
t— LU
'ro
,2
r—
ro
•4-^
0
i—
+j
s_
o
0.
to
c
ro
S-
1—
cn
c:
•>^
OJ
^
i
>>
ro
C
O
•r-
| *
ro
4->
CO
to
ro
C3
^^
•i—
O
i —
(O O
-)
•fJ
O
s_
4->
U
QJ
n—
LU
**s^
~^
^
t:
o
j->
Q.
E
to
c:
o
c
o
•r—
-U
fO
1 —
c^
o
f—
•r—
O
r-*-
ra
o
(_>
>,
1o
a>
31
*»_^«'
c
o
(/)
s_
0)
a.
13
^;
rn
O
r-~
"wT
-a
c
ro
to
3
O
f^-
-!->
(_J
O
*s
1—
CO
•—
^~
•
CO
n
«
r—
LO
.
o
co
CM
LO
0
O
U3
LO
U~>
,
O
CM
CD
i —
O
UD
^^
LO
^J-
^r
LO
CO
1
, —
(-^
cn
'
c
ro
-D
LO
•—
!
*
p^
^j.
(
,_
r__
.
CD
CM
00
IO
O
cn
CO
^J-
,
CD
1 —
.
CO
o
cn
CO
LO
•c^-
CD
,
CM
1
r«
f^^
CTv
(_)
CD
Q
CO
CO
CO
^^
CO
.
CM
O
^_
r~
CM
t--
r.
CD
CO
^3-
.
CD
^_
.
^~
r—
f-^
CM
1 —
LO
UD
CM
CM
CM
1
LO
f-s^
cn
>
o
•z.
o
CM
O
•
CO
UD
r_
^>
,
0
CO
CD
*— •»
,
,
CD
UD
CM
O
O
LO
LO
CM
CM
LO
*^-
nf
^J
j—f
t^.
CTV
1 —
ai
c
3
•-3
O
CO
cn
CM
CM
•
CM
LO
f^^.
.
O
LO
r~-
CD
f-^
0
o
CO
t~-
co
,
o
cn
LO
CO
CO
CO
LO
LO
-*
(^
1~~
LO
r-
cn
i —
O)
c
•^
•o
106
-------
TABLE 16. BREAKDOWN BY POINT OF RELEASE OF ENERGY USED IN THE FAIRBANKS
AREA. ALL FIGURES IN MILLION KWH PER MONTH OR YEAR
1973
T
F
M
A
M
>)
J
A
s
0
N
1)
Y
1974
J
F
M
A
M
J
J
A
S
0
N
D
y
1975
J
F
M
A
M
J
J
A
S
0
N
D
Y
Oil
9
3
2
3
10
14
16
17
14
17
(>
(>
117
4
6
20
9
12
20
23
21
27
27
24
32
224
17
23
32
33
30
35
30
37
44
46
48
30
405
Mobile
Gasoline
38
26
41
43
49
49
61
57
60
47
4')
52
572
45
44
40
46
52
59
65
70
63
54
52
49
640
51
54
70
80
82
104
113
114
113
112
no
90
1093
Dispersed Stationary
Total
47
29
43
46
59
63
77
74
74
59
55
58
684"
49
50
60
55
64
79
88
91
90
81
76
81
864
68
77
102
113
112
139
143
151
157
158
158
120
1498
Oil
98
70
56
3R
27
16
18
26
31
52
73
87
W
97
100
73
47
27
25
22
35
38
68
80
98
710
129
98
77
67
35
35
31
38
55
80
125
126
896
Healy
Elec.
17
15
10
15
14
12
13
14
10
13
I/
17
T66
17
16
16
12
15
14
14
15
11
18
18
17
184
19
17
18
17
12
16
16
15
12
18
18
18
194
Local
Elec.
22
16
19
9
7
7
6
6
12
14
16
14
147
18
17
14
11
8
7
7
7
12
13
20
24
158
27
21
22
14
14
10
10
13
16
22
33
31
233
UofA
15
12
12
7
5
4
3
5
7
12
15
15
112
16
15
13
8
6
4
4
4
5
11
13
15
114
16
13
12
9
6
5
5
5
12
14
14
15
126
Concentrated
Military
(power plant waste heat)
Total
152
113
97
69
53
39
40
51
60
91
121
133
1017
148
148
116
78
56
50
47
61
66
110
131
154
1166
191
149
129
107
67
66
62
71
95
134
190
190
1449
Oil
31
15
19
1
0
0
3
4
10
12
18
20
"135
20
23
7
4
1
0
1
9
7
6
18
31
128
52
31
31
16
14
3
3
14
24
34
59
54
335
MUS
Coal
33
28
29
26
17
18
17
18
23
24
29
27
291
31
25
30
24
20
18
17
17
24
27
33
37
303
33
29
30
25
24
23
22
21
23
28
34
39
331
Total
66
43
47
29
17
19
22
25
33
35
49
50
434
52
49
37
28
21
20
17
28
31
32
51
67
435
87
59
61
42
37
26
24
35
47
62
92
95
667
75
61
64
48
45
32
32
37
47
60
65
76
"6*2"
77
75
69
47
38
28
37
39
43
67
70
96
686
100
87
77
57
52
41
32
36
50
51
76
90
749
107
-------
The military plants are heating as well as power plants, and their total
output was simply calculated at 3500 KWH/ton of coal and 38 KWH/gal of fuel
oil, fuel oil contributing less than 1 million KWH in most months.
During 1975, the energy use peaked at 519 million KWH per month in
November (=720,000 KW or 1.03 x 10 cal/min). The minimum energy use, in an
unusually warm July, was 264 million KWH per month (=355,000 KW or 0.51 x
10 cal/min). For practical computational purposes, the energy consumption
in Fairbanks may be set at 10 cal/min in a cold winter month and 0.5 x 10
cal/min in a warm summer month.
PRODUCTION OF UNAVOIDABLE COMBUSTION PRODUCTS
Water vapor (H?0) and carbon dioxide (C02) production are dependent on
fuel consumed. Benson (1970) calculates that 1.38 kg of H20 and 3.10 kg of
C02 are produced for each kg of gasoline burned, 1.33 kg of HpO and 3.13 kg
of C02 are produced for each kg of fuel oil, and 0.68 kg of H20 and 2.36 kg
of C02 are produced for each kg of coal (as delivered). As a gallon of
gasoline weighs 2.99 kg, a gallon of fuel oil weighs 3.36 kg and a ton of
coal weighs 907.2 kg, the H?0 and C02 produced by complete combustion may be
obtained by multiplying the numbers in Tables 8, 9, 10, 11, and 13 by the
numbers in Table 17. Table 18 shows some representative calculations of C02
and H?0 for 1975. These values do not include H^O release from activities
other than fuel consumption.
TABLE 17. PRODUCTION OF H?0 AND C02 PER UNIT OF FOSSIL
FUEL BURNED (COMPLETE COMBUSTION ASSUMED)
Fuel
Coal
(tons)
Gasoline
(thousand
gallons)
Fuel Oil
(thousand
gallons)
H20 (thousand kg)
.617
4.13
4.47
C02 (thousand kg)
2.14
9.28
10.52
100
-------
TABLE 18. H20 AND C02 (IN MILLIONS OF KG/MONTH) GENERATED
BY FOSSIL FUEL COMBUSTION IN THE FAIRBANKS AREA
IN 1975. MONTHLY TOTALS EXCLUDE COAL HEAT.
Mobile
J
F
M
A
M
J
J
A
S
0
N
D
H20
8.1
9.1
12.1
13.4
13.4
16.5
16.9
18.0
18.5
18.7
18.7
14.3
co2
18.4
20.8
27.5
30.4
30.4
37.4
38.3
40.9
42.2
42.6
42.6
32.4
Oil
Heat
Coal
LI A pr\ u r\
15.2
11.5
9.0
7.9
4.1
4.1
3.6
4.4
6.5
9.4
14.7
14.9
3b.
27.
21.
18.
9.
9.
8.
10.
15.
22.
34.
35.
7
1
2
6
8
7
5
4
2
1
6
0
Heat Power Plants
C02 H
36
29
27
20
18
14
12
14
20
23
33
36
Total
20 C02 H20
.1
.4
.7
.2
.2
.2
.4
.7
.1
.6
.2
.4
115
96
90
67
60
48
42
48
65
75
104
116
.9
.5
.8
.2
.8
.7
.7
.5
.5
.8
.9
.8
59
50
49
41
36
35
33
37
45
52
67
66
co2
170
144
140
116
101
96
89
100
123
141
182
184
178
404
105 248
30
286 935
578 1616
109
-------
SECTION 9
THEORY
INTRODUCTION
There are several empirical relationships between heat island intensity,
AT , and the observed meteorological and geographical setting (Landsberg,
max , , . i ,p
1974). Oke (1972) suggested two relationships; first, AT av = * ' /4v ,
max
where is population and v is mean wind speed. He also suggested that the
maximum heat island in the case of no wind and no cloud cover is given by
AT = 2.96 log - 6.41. Ludwig (1970) placed more emphasis on lapse rate
* 1 /d HT
in the first 100 m, with the equation A! „ = ' (.0633-. 298 ^-) where
i-p max op
j~ is the lapse rate in °C/mb. If all three equations are applied to the
Fairbanks area (population 65,000) with an assumed temperature of -20°C, a
surface pressure of 1000 mb, a 100 m inversion of 15°C (^- = -1.11) and a
_1 ap
wind speed of 1 m sec" we obtain:
ATmax =4-°°C (0ke' v = 1}
AT ,„ = 7.8 (Oke, clear and calm)
max
ATmax = 6<3°C (Ludwi9)
The Fairbanks heat island observed in the present study with clear night
skies, snow cover and light winds has very rarely been less than 10°C, and
has on occasion been measured as 14°C. This discrepancy may be attributed
to the fact that the combination of lapse rate, wind speed and per capita
heat production at Fairbanks is completely outside the range for which the
*
This refers to the difference in air temperatures measured at the ground
and at an altitude of 100 m. The actual air temperature profile is often
complex and involves stronger gradients in places as discussed later in the
text.
110
-------
empirical models were developed. Indeed, these factors combined with the
absence of solar radiation in winter make Fairbanks an ideal place to study
certain aspects of the urban heat island.
Computer simulations have been carried out in considerably more detail
(e.g., Leahey and Friend, 1971, Atwater and Pandolfo, 1975). However, these
do not generally give the form of the dependence on such variables as
population, wind speed and lapse rate, which is what we intend to do here.
The approach used is similar to that of Leahey and Friend (1971).
Physically, the intensity of a heat island is determined by heat
addition to the air by the substrate (which is strongly influenced by how
long the air stays over the city, i.e., wind speed), by how this heat is
distributed vertically and (if the air contains gasses or aerosols active in
the thermal infrared) by increased radiative energy loss from the warmer
air. Increased radiative loss from the substrate will influence the amount
of heat available for transfer to the air, but will also be influenced by
the substrate temperature, which is tied to the air temperature. We assume
that in Fairbanks in winter, with very low hLO content of the air, thermal
radiation from the air need be considered only if fog (ice fog) is present.
At zero background wind speed, no fog, and a fixed total energy loss from
the substrate, the intensity of the steady-state heat island will be de-
termined by a balance among three factors: Higher infrared radiative loss
from the warmer city substrate (buildings, pavement, automobiles, and air-
warmed natural surfaces), increased convective energy transfer from the
substrate to the air, and air exchange with the surroundings induced by the
horizontal temperature gradient around the city. If fog, other aerosols or
gases with strong infrared absorption bands are present in quantity, the net
radiative loss from the substrate will be partially or wholly absorbed by
the local air. The upper part of the optically active layer will then re-
radiate this energy away from the city. If the temperature at the top of
this layer approximates the substrate temperature (as will normally be true
for thin layers) the total radiative energy loss will be the same with or
without fog, even though the mechanics of the energy loss differ somewhat.
The background lapse rate, provided it is stable, would affect the length of
time required to reach equilibrium but should have little effect on the
equilibrium temperature in the zero-wind case. The zero-wind case, combined
111
-------
with fog, but neglecting the heat island circulation, was considered by
Bowling (1970).
The case with wind but neglecting radiative cooling has been considered
by Summers (1965). The basic assumption is that stable air moving over the
urban surface is heated uniformly from below. The heating results in an
adiabatic lapse rate reaching from the ground up to some height Z at which
it intersects the undisturbed temperature profile. The amount of temperature
change is determined by the requirement that the change in the thermal
energy of the air column,
fz
AT(z)pc dz, be equal to the energy added by the heating from below. (Here
•'o p
z is the vertical coordinate, p is the density of air and c is the specific
heat at constant pressure of air.) This approach actually overestimates the
heat island intensity, as it does not allow for the increased radiative
energy loss from the warm city surface. Increased radiative loss from the
air is neglected for the moment. Application of the Stefan-Boltzman law
_2
shows that at -20°C, the radiative energy loss increases by .5 x 10 cal
-2 -1
cm min for each degree Celsius temperature increase.
In general, the input to the steady-state nighttime heat island is a
fixed energy flux, h. The substrate temperature, T, at any point is deter-
mined by how this energy flux h, is partitioned between a part, q, which heats
the air, and a part lost through radiative cooling. For any particular
energy flow, h, from the substrate, there will exist an ideal radiative-
equilibrium heat island intensity, ATre, at which the increase in radiative
loss over the city relative to the background is just balanced by the energy
flux, h, through the substrate. T = T + AT is the radiative-equilibrium
temperature at a point. It must be emphasized strongly that AT and T
are limiting values which will be observed only under certain transient
conditions (e.g., a local temperature maximum along a flow line). T = T
implies q =0, and if this condition held area-wide in the presense of any
cross-isotherm wind component, no heat island could exist. The energy
balance for the city may be written as
h • " - »T4'
112
-------
TQ is the background temperature, assumed to be in radiative equilibrium
with the sky; i.e., TQ = Tre for h = 0, and o is the Stefan-Boltzmann
constant, a - 8.132 x 10"11 cal cm"2 min"1 °K~4. The relationship of Tpe
to the observed temperature T, to q and to h is shown in Figure 53.
Normally the energy flux h will vary through a city. For any given
point, then, an ideal radiative-equilibrium temperature T is defined from
equation (6). T is the temperature at which h is entirely radiated away,
and no energy is available for transfer into the air column at that point,
* dT
i.e., the Lagrangian derivative of the air temperature, ^rr, would be equal
to zero. If the air temperature at y is less than T (y) (as will normally
be the case when h is increasing along the wind) convective transfer will be
active and may consume a substantial portion of the available energy (q > 0).
If the air temperature at y is greater than T (y) (as may be the case where
h is decreasing along the wind) radiative cooling will occur at the surface
(q < 0). The difference between T and T , i.e., AT , the radiative-
equilibrium heat island, may be obtained by expanding oT = o(T +AT )
I c U ic
as a polynomial: aTre4 = a TQ4 + f TQ2 ATre + 6TQ2(ATre)2
-> A AT
+ 4 TQ(ATre) + (ATre) . It is then apparent that if -y-^ «1,
aTre4 = A To4 * 4 TQ3 ATre, so oTre4 - aT04 ~- 4 1^1^. From (6) we
then have
ATpe £ h/4a TQ3 (7)
The maximum temperature measured in a particular heat island situation in
any traverse along the mean wind should satisfy equation (7) for the value
of h at that particular point. Figure 54 is a cross section through a
hypothetical city in which h increases to a maximum and then decreases. The
wind is assumed to blow from left to right over a city with its center and
maximum heat input at y =R. Note that the core of the heat island is down-
wind of the maximum city heat input, and that q = 0 and AT = ATre at the
heat island core.
An expression for the heat actually transferred to the air, q, may be
obtained by using equation 7. To a good approximation,
*i.e., the derivative calculated moving with the air column.
113
-------
incoming
radiation
outgoing
4 radiation( __ 4
Tre=T0
h = 0
B
outgoing
radiation
heat
'conducted^—> d_T
to air "q dt
(
-------
q = h - 4aT3AT, (8)
where AT is the difference between the temperature T measured at any point
in the city, and the background temperature T . Note again that this AT is
normally not AT , and that if AT = AT , q = 0.
Note that h includes all of the added heat due to the presence of the
city - not only that added by human activity, but also that due to albedo
differences, differences in the rates at which various substances respond
to changes in their energy balances, and differences in the heat loss by
evapotranspiration. Only changes in roughness are not included. Our
calculations for Fairbanks will include only the man-made heating, because
the other terms are believed to be small in Fairbanks in midwinter. In most
climates they must be included.
In addition to neglecting the radiative effect, the previously mentioned
empirical approaches neglect the possible influence of the form of the lapse
rate. Thus a temperature profile which is isothermal to 90 m and capped by
an inversion of 3° in 10 meters would not be expected to have the same
effect as a 3° inversion in the lowest 10 meters with a 90 m isothermal
layer above, even though both give identical 100 m lapse rates.
The purpose of this section is to consider theoretically two extensions
of the Summers approach. The first to be considered will be the effect of
the form of the lapse rate in the case where q, rather than h, is assumed to
be known. Four types of temperature profiles will be considered: a simple
linear profile, a logarithmic profile (inversion strength decreasing with
altitude), a step function with sharp inversions separating regions of
adiabatic lapse rate, and a capping inversion. The second extension involves
including the radiative loss for a specific but not unreasonable case in
which the equations can be handled analytically, followed by an examination
of the functional dependence of the heat island intensity (measured at the
city center) on population, population density, wind speed, inversion
strength, and energy use per capita. The particular case considered is one
in which the value of h increases linearly toward the center of a circular
city and the initial lapse rate is constant with height. Many of the
assumptions are similar to those of Leahey and Friend (1971), but they are
applied to yive an analytical result which allows consideration of the
115
-------
expected heat island as a function of such factors as population density.
All calculations are valid only upwind of the maximum heat island,
i.e., for q > 0 at all points. The conditions for the radiative partition
case are even more rigid, as this calculation is valid only for h increasing
linearly. Downwind of the point at which q becomes 0 (i.e., where AT =
o
h/4oT ) if h is decreasing the substrate will be too warm for h to balance
the radiative loss. If the air is clear the substrate will cool radiatively;
if the air is optically active the radiation will take place from the mixed
layer, as shown in Figure 54. In either case the ground level temperature
will decrease downwind of the point at which q becomes 0.
The effect of the circulation induced by the heat island on the strength
of the heat island is neglected throughout, although it may have considerable
influence at very low wind speeds. Another assumption which must be kept in
mind is that the lapse rate in the mixed layer is adiabatic, with the tempera-
ture at the base of the mixed layer being equal to that of the substrate.
In actual fact, the substrate must be slightly warmer than the air and the
lapse rate must be slightly superadiabatic for convective heat transfer to
be possible. We here assume that the difference from the adiabatic case is
small relative to the difference between the initial lapse rate and the
adiabatic one. Thus the approach taken here is not applicable to initial
lapse rates near the adiabatic.
Still another possible source of error is the assumption that the
background temperature, T , is the radiative-equilibrium temperature. Heat
island intensity is normally measured at 2 m, but the 2 m background tempera-
ture may be noticeably greater than that at the ground surface (up to about
5°C). As it is the ground surface which approaches, but normally does not
reach, the radiative equilibrium temperature, the heat island strength
predicted by the model will be slightly overestimated.
The calculations to be made assume partition of h into two fractions at
the city surface. The radiative fraction is emitted directly from the
substrate and lost to the system, while the conducted fraction, q, is con-
ducted to the base of the atmosphere and then mixed convectively. Absorption
and re-radiation by the atmosphere are considered negligible, and in the
thin boundary layer being considered, this is a good approximation for clear
air (i.e., with normal amount of CO,, and H^O as the only optically active
116
-------
Observed AT
0
c.
fel
9
Maximum
radiative cooling (
from substrate }j
o>
a>
X
0
d.
I
mixed layer with fog
I
I / radiative cooling
I \ from fog
y=R y=Y
Figure 54. Cross section through a hypothetical city, wind from left to
right in all cases. Panel a: asuumed heat input, h, and heat
transferred to the air, q. Panel b: values of ATro and observed
AT for the given values of h and q. Panel C: height of mixed
layer, Z, and development of inversion downwind of maximum heat
island under clear conditions. Panel d: height of mixed layer, Z,
and cooling downwind of maximum heat island under foggy conditions
117
-------
components). In turbid air, and particularly in the extreme case of optically
thick fog in the city with clear air in the background areas, the energy
radiated from the substrate will be partially to totally absorbed in the
lower part of the mixed layer. This excess energy will be transferred to
the fog top by convection, where it will be radiated away. If the fog top
(assumed to be at the top of the mixing layer) were exactly the temperature
of the substrate, the net radiative loss from the city surface/mixing layer
system would be unchanged, and thus the net heat addition to the system
would be unchanged. In actual fact, a fog layer of finite depth heated from
below must be slightly cooler at the top than at the bottom, as an adiabatic
lapse rate becomes established in the fog. Because the top of the fog layer
is cooler than the substrate, the radiative loss from the fog top is less than
it would have been without the fog. This reduces the fraction of h which is
lost by radiation and inceases q, the amount retained in the city air; the
net result is an increase in the heat island intensity AT. However, we will
neglect this affect and assume that the radiative loss from the city, and thus
the energy available to heat the air,is the same with or without a fog.
In addition, the modification of the background lapse rate proceeds a
little differently in the presence of fog. Figure 55 shows corresponding
steps in the process of modification with and without a fog in the mixing
layer. In each case the area on the diagram between the original and
modified sounding is proportional to the added heat, and the heat added
in the two cases is approximately the same. However, in the foggy case
radiative cooling at the top, Z, will lead to a temperature below that of the
unmodified sounding. In order to have the same energy gain in both the clear
and the foggy cases the layer which suffers a net energy loss (horizontally
shaded) must be balanced by increases (heavily vertically shaded) in the
layer which enjoys a net energy gain (vertically shaded). This can only be
done within the constraints of an adiabatic mixing layer by allowing the
adiabatic layer to have a higher temperature. The foggy city is then
predicted to have a higher temperature, a deeper mixing layer, and stronger
vertical overturning than the clear one. This applies when the background
area is not foggy.
Inability to find a fog-free area upwind of and at the same altitude as
the city of Fairbanks is probably the primary reason why measurements of the
118
-------
Fairbanks heat island showed the foggy heat island to be consistently less
than the heat island under clear skies.
THE EFFECT OF THE FORM OF THE TEMPERATURE PROFILE
We now proceed to examine the effect of different forms of the lapse
rate on the intensity of the heat island. For this purpose we will define a
standard Cartesian coordinate system with z being the vertical coordinate
Background
Sounding
a>
I
Zf
Potential Temperature
Figure 55. City vs. background soundings for clear and city-fog cases. Both
city soundings have had the same net heat addition.
119
-------
and y being the coordinate along the wind, with the y axis running through
the city core and the origin of the system being the upwind edge of the
city. In order to simplify the calculation we will use the potential temper-
ature e(y,z) (neglecting the x coordinate), e and T are related by e(y,z)
= 0 + T(y,z) + .01 z, where z is in meters, T is in °C, and .01 is the
adiabatic lapse rate, .01 °K nf . For our purposes, e is arbitrary, as we
will be interested only in differences in e. The heat island intensity at y
is then g^en by AT = T(y,0) - T(0,0) = e(y,0) - e(0,0) = A6. Figure 56 and
Table 19 are included to aid in the definition of the symbols to be used.
Upwind
Potential
Temperature
Profile
(y
-------
TABLE 19. SYMBOLS USED
Variables:
8o2T 6 , , _2
/\ = (energy length" temp" time" )
PCP
2
B = r/v (temp length" time)
c = specific heat of air at constant pressure
E = total energy released by heating
, energy
surface
h = h(y) = h(x,y), energy flux (energy area" time" ) from the city
2
G = irR /cj>, area per person
P - atmospheric pressure
q - q(y) = q(x,y) conductive energy flux from the city surface to the
air
/•y , >
Q(y) - I 3.1X1 (jyj tne energy accumulated by a unit column of air traveling
J pcpv from y = 0 to y, divided by the volume heat capacity
of the air.
T = temperature, usually in °K
T = background temperature
T a = temperature at warmest part of heat island
max
T = radiative equilibrium temperature (eq 1)
AT - T - TQ
R = radius of city
v = mean wind speed
x = horizontal coordinate perpendicular to the wind
continued
121
-------
Table 19 (Cont.)
y = horizontal coordinate along the wind
Y = y-coordinate of the point of maximum temperature in the heat isalnd
z = height coordinate
Z = Z(y) = mixing height; Z(0) = 0
deQ
Y = y(z) = potential lapse rate = T~—
e = G + T(y,z) + .01 z = potential temperature
0 = for our purposes, an arbitrary constant
p = air density
= population
Constants:
a = lapse rate at z = 0 in logarithmic profile
b = constant in logarithmic profile
C = height of base of capping inversion
J = rate of change of h along the wind
K = rate of change of q along the wind
m = ratio of heat island intensities with and without fog
z. = layer limits in step function temperature profile
r = lapse rate for the linear temperature profile
Y- = lapse rate in layer i in the step function and capping inversion
1 temperature profiles
o = Stefan-Boltzmann constant, = 8.132 x 10 cal cnf min~ °K~ .
122
-------
O(y,z) = 6(0,Z(y)) 0 < z < Z(y)
(9a)
o(y,z) = o(0,z)
Z(y)
(9b)
Z(y) is the mixing depth, and is determined as follows:
The wind is assumed to be along the y axis with speed v. The rate
of heat release to the air per unit area by the substrate is q(y), which
gives an amount of heat addition ^^' ^ to a column of air with a unit
basal area as it travels a distance dy. If Z is small enough that the density
of air, p, does not vary too much with height, the energy per unit area
involved in changing the height of the top of the mixed layer from Z(y) to
de
dZ(y),
Z
Z(y+dy) is Pc Z(y)de , where dev = e(0,Z(y+dy)) - e(0,Z(y)) =
v y y
and c is the specific heat of air at constant pressure. If this is set
equal to the added heat, ' y , then
dz
PcpZdZ -
(10)
or
dz
ZdZ
- q(y)
pcpv
Ay.
(ID
Since Z = 0 at y = 0,
de.
and Y(Z) = —^
dy.
(12)
(13)
(14)
123
-------
Then
Z Y(Z)dZ = Q(y).
(15)
We will consider four cases:
Case 1: linear profile
OQ(Z) = oQ(0) + r z; Y(Z) = r = const > 0
Q(y) =
z(y) =
r ZdZ = r Z
(16)
(17)
(18)
The ground potential temperature = 6Q(Z) = 9Q(0) + rZ, so
AT = oQ(Z) - oo(0) = rZ =
Case 2: logarithmic profile
e0(z) =f In (l+bz)+oo(0); Y(Z) =
(19)
(20)
(Physically, a is the initial lapse rate at z = 0; b is the inverse of
the height at which the lapse rate is one half of that at the surface.)
Q(y) =
a/. H7 - ££ + £_ />„ r_J ^
1+bZ L ~ b b2 *-n M+bZ;
(21)
In this case an analytic solution for Z, and thus for AT, is not
practical. Results for a particular numerical case will be discussed
below.
124
-------
Case 3: step function profile
eo(z} = e0(0)
z—zl
Y(Z) = YI
Y(Z) = 0
Y(Z) = Yc
Y(Z) = 0
Y(Z) = Yt-
z,, Zp, etc., are heights of the tops and bases of the adiabatic layers.
For Z F zp Z YI dZ = Q(y) (23)
(22)
(24)
up to the y at which Q(y) =
At that point I makes a discontinuous jump to
vz2 f
Then Q(y) = J + ^3Zclz
,2 , 2
-) until Z = z,,
125
-------
where Q(y) =
, etc.
Between z2 and z3, for example.
«/ .
Q(y) =
222
Z Y Z Y Z
or
= Q(y) -
and
Z =-
(26)
The heat island intensity, 0(0,Z) - e(0,0) = AT is given by
AT = AO = Y1Z1 + Y3 (Z-Z2).
(27)
Case 4: capping inversion
o0(z) = oQ(0)
YI
0
-------
(30)
and AT = A6 = Y2Y-|Q(y) • (31)
For Z > C,
Z
2' o
fZ Y? y ~
±- + Y2ZdZ = Y] £- + -f- (ld-Cd) (32)
C
2Q(y)+C2(Y?-Y,) 2
—— (valid only for Q(y) > Y, —-, which (33)
Y2 is automatic for Z >C)
and AT = A6 = C(Y,-Y9) +Hfro[2Q(y)+C (Y9-Y,)]. (34)
I L. \ C £. I
In order to compare the four cases with each other, with the Fairbanks case
and with the empirical inversion model of Ludwig (1970), calculations were
carried out for each of the four cases with the constants adjusted to give
the same 100 m inversion for all four cases. The constants used were:
Case 1 r = .1522 °K m"1
Case 2 a = 1 °K m"1, b = .2 m"1
Case 3 z-j = 10 m, z2 = 45 m, z^ = 55 m, z^ = 89.56 m
Y = Y3 = Y5 = -5 °K m"""
Case 4 Y] = .01 °K m"1, Y£ = 1.432 °K m"1, C = 90 m
All four cases give a 15.22°K potential temperature inversion (= 14.22°C
measured inversion) over the first hundred meters of the sounding. (This is
quite a large inversion, but by no means unknown in Fairbanks, as shown in
the discussion below. The use of .1522 is to give simple constants in the
logarithmic case.) In addition, calculations were carried out for a 100 m
*
potential temperature inversion of 4.6°K (3.6°C/100 m) for cases 1 and 2.
Case 1 r = .406, case 2 a = 1°K nf1, b = 1 m"1
127
-------
Potential Temperature Difference from Surface Temperature
Figure 57. Potential temperature profiles for which calculations were made
of mixing height and heat island intensity. Light lines - lapse
rates with the 100 meter potential temperature difference equal
to 15.22°C. Plain line - linear case; line with crosses - log-
arithmic case; dashed line - stepped case; dotted line - capping
inversion case. Heavy lines - linear and logarithmic profiles
for 100 meter potential temperature differences of 4.6°C (3.6C/
100 m inversion).
-------
The various initial soundings used are shown in Figure 57; the resulting
heat island intensities and heights are shown in Figures 58 and 59. Note
that the form of the sounding is of sufficient importance to outweigh a
three-fold change in the lapse rate in some cases. A good deal of the
scatter in Ludwig's (1970) plots is probably due to this kind of variation
in the form of the lapse rate.
In order to compare these calculations with the Fairbanks case, we will
consider two cases in which Fairbanks helicopter data were available -
traverses 24 (14 March 1975) and 38 (26 Feb 1976). Table 20 gives the
important temperatures for both cases.
I40r
200
400
600
800
Figure 58.
Q(y), °Km
Mixina height,'Z as a function of accumulated bottom
heating Q(y), in °K m. Line types as in Figure 57.
Short double lines indicate the 100 m level.
129
-------
0
200
800
400 600
Q(y), °Km
Figure 59. Heat island intensity, AO, as a function of Q(y). Line types
as in Figure 57. Short double lines indicate where the heat
island extends above 100 m.
TABLE 20
TEMPERATURES FROM TWO HELICOPTER TRAVERSES
Upwind
Ground
Terno.
at y = 0
#24 -17°C
#38 -31 °C
City Heat Upwind 100 m
Core Island 100 m Potential
Temp. AT Temp. Temperature
y = y at y = 0 Inversion
z=2m z=100m
-9°C 8°C -3°C 15°C
-18.5°C 12.5°C -16°C 16°C
Geo. Form
Inst. of
Temp. Lapse
Rate
-9°C linear
(heli-
copter
ascent)
-18°C ^ log-
arithmic
(tethered
balloon
1st 10 m)
13U
-------
If the observed heat island intensities of 8°C and 12.5°C are located
on the linear and logarithmic plots, respectively, both are found to corres-
pond to Q values of slightly more than 200. This value of Q gives mixing
heights of about 60 m in each case. Measured temperatures at the Geophysical
Institute, 60 m above the city, are in good agreement with the observed city
temperatures, which aids in confirming the original lapse rate assumptions.
CONDUCTIVE/CONVECTIVE AND RADIATIVE ENERGY LOSSES
There are some problems in direct calculation of Q(y). The value of
q(y) along a particular path is not generally known, and in addition, q(y)
is actually dependent on the temperature at y (see eq. 8) - a fact which we
ignored in the solution of our initial equation. In the case of Fairbanks,
the mean wind speed is also poorly known, as most of the measured winds
probably represent eddies and waves (Section 7). It is, however, possible
to set some limits to q and to v with the available data. To do so, we will
consider the role of radiative loss.
Let h(x,y) be the total energy per unit area supplied at the surface by
the city. This will be the sum of q(x,y), and the net radiative energy
loss. If the outlying areas are in radiative equilibrium at temperature T ,
and the incoming longwave radiation is the same in and out of the city, then
the net radiative energy loss is aT (x,y) - aT = 4aT AT(x,y) (where AT
is the heat island intensity at (x,y)) assuming that the average substrate
temperature is close to that of the overlying air. Then
h(x,y) = q(x,y) + 4aTQ3 AT(x,y). (81)
Since it is necessary to know q to calculate AT, while h is the actual
input, this equation is of use only in special circumstances.
One such special circumstance is the case in which q increases linearly
along the path of the wind to the city center and the lapse rate is a linear
inversion, ~ = r. Let q(0,y) = Ky. Then from eq (23),
oZ
*"' (35>
131
-------
and from eq (19), AT = A6Q = /2rQ(y) = y
(36)
Eq (81 ) then becomes
If we set
h(0,y) = Jy or J =
in parallel with
q(0,y) = Ky or K =
x=0
(37)
(38)
(39a)
x=0
(39b)
Thus for a linear increase in h along the wind and an initial uniform
inversion, we also have a linear increase in q(0,y). If J is known, K can
be calculated from the defining equation for J, squared:
From this,
K2 - 2KJ + J2 = 16 a2T 6 rK
o pcpv
9 I 8 0T r \
K "2K + ~~
(40)
(41)
Let
and
= A
(42)
(43)
132
-------
Eq. 38 demands that K < J, which allows only the solution (Dwight, 1961,
eq. 55.1)
K =
J + AB + /2ABJ+(AB)2
(44)
AB + /2 AB + (AB}2
J J d
(45)
The fraction of the heat energy released by the city which is actually
heating the air is £, which now can be written as:
K
J
AB
J
AB 4. /AB^2
J 1J >
(46)
AB
Values of K/J for different values of y- are given in Table 21.
TABLE 21
K/J AS A FUNCTION OF AB/J
AB
J
.001
.01
.02
.05
.1
.25
1
3
5
10
K/J
.96
.87
.82
.73
.64
.50
.27
.13
.08
.05
133
-------
The three variables A, B and J represent the regional meteorology, the
local micrometeorology and the city heat addition, respectively. Using the
gas law to express air density in terms of temperature, T, and pressure, P,
A = 6.327 x 10"16 T7 P"1, (47)
-1 -1 -?
where T is in °K, P5 in mb and A in cal cm °K min .
-1 -1
If r is expressed in °K m and v in m sec , then
-4
B = ~= 1-667 x 10"6 in °K min cm"2. (48)
Some values of A, B and AB for meteorological variables typical of
temperate-zone cities and of Fairbanks are given in Table 22.
TABLE 22
TYPICAL VALUES. UNITS AS GIVEN IN TEXT
T
20°C
10°C
-20°C
-40°C
P
700
1000
1000
1000
A
1.68x10"'
9. 20x1 O"2
4.20xlO"2
2.4xlO"2
r
.01
.01
.15
.15
V
5
5
.25
.25
B
3.33xlO"9
3.33xlO"9
l.OlxlO"6
l.OlxlO"6
AB
5.59xlO"10
3.07xlO~10
4.26xlO"8
2.44xlO"8
Note that r = ~p = .01 is the isothermal case (^ = 0) and that the
calculations for r = .15 actually refer to r = .1522 as used in the discussion
of the influence of the form of the lapse rate.
If the total energy, E, released by the city is known, J may be estimated
by assuming a circularly symmetric city of radius R. Let the heat release,
h, be zero outside the radius R and increase linearly toward the city center.
Then
h(x,y) = J R - v£2+(R-y)2 for x2 + (R-y)2 < R2 (49)
h(x,y) = 0 for x2 + (R-y)2 >_ R2.
134
-------
This defines the conical heating distribution used from here on. The total
heat release, E. is the area integral of h(x,y) over the city. By defining
2 2
r = /x +(R-y) , this integral is easily evaluated, giving
E = - (50)
or J = - . (51)
TrfT
For Fairbanks and its surroundings in winter, we have from section 8
E = 1010 cal min"1.
If R = 5 km, appropriate for the area as a whole, and we assume values
for the meteorological variables as given in line 3 of Table 22,
3 x 10 7>64 x 1Q-8
>
Tr(5 X 10V
giving AB/J = .56, or K/J = .36 - only about 35% of the energy released goes
into heating the air. (If a dense fog is present, the 35% is what is left
over to heat the air after subtracting the radiation from the fog top.) If
R = 1 km, which might be more appropriate for air coming into the city from
the north,
J = 9.55 x 10"6, AB/J = 4.46 x 10"3 and K/J = .91.
90% of the energy goes into heating the air in this case.
Note that the above analysis and the definitions for K and J hold only
upwind of the city center. Downwind of the maximum value of h at the city
center, AT will continue to increase as h(0,y) decreases until q of eq (81)
q
becomes zero, i.e., 4aT AT = h(0,y). At the point where q = 0, let y = Y.
Since from eq (36) AT is proportional to y and thus to h = Jy up to y = R, q
3
= h - 4aT AT, and AT continues to increase while h decreases from y = R to
y = Y, q must decrease much more rapidly than it increased, becoming 0 at Y
and negative thereafter (Fig. 53). Negative q implies heat transfer from
the air to the ground, so the temperature and thus AT decreases with y for y
> Y. Since AT increases when y < Y, and decreases when y > Y, it follows
that Y is the coordinate of the maximum heat island intensity. The maximum
135
-------
heat island intensity, AT(Y) must be within the city, i.e., Y must be less
than 2R. This follows from
x=0
_
dy
= K
for y < R,
h(0,y) = JR-(y-R)J, from eq (49)
- 4aTAT)
= -J - 4- (4aT3AT).
definition of Y,
dy
d_
dy
For R 0 by the
so -
dy
> J > K =
dq
R_ R
and that AT (R) > AT,, . Te er iit t the he islnd inensit is
re — max
* f\ t tll\A1ll I I\~\A \* I *J I U I I M I I I W*_> I I o^ * \*J IliU r «-"_- *~r I \AN*I^N*WV>^« "J I I vx ** I I I ^ s* i < w* w i
UMM ^,,ul, ^T^(R) > ATm,u. The upper limit to the heat island intensity
re — max
given by AT , = AT JR). This can be determined from
max re
h(y)
= 4aT3ATre = RJ
or AT
y=R
= RJ
6 4oT,
3E
(52)
(53)
136
-------
For the two Fairbanks cases discussed above, this gives
for R = 5 km ATre = 7.2°K
for R = 1 km ATre = 181°K.
Obviously the upper limit is a very poor approximation to the 1 km case,
and in fact the assumptions in this case are so extreme that the AT/T «1
assumption breaks down. A real energy concentration of this magnitude (720
-2 -2
W m at the city core; compare with an estimated 630 W m for Manhattan
Island, SMIC, 1971) would also almost certainly invalidate our assumption
that the circulation induced by the heat island was small.
The actual heat islands predicted at the city core, which are also the
lower limits for the maximum heat islands, using the figures listed and eq
(36) are:
TABLE 23
PREDICTED HEAT ISLANDS
T
°C
-20
-20
-20
-20
P
mb
1000
1000
1000
1000
r _, v _,
°C m~ m sec
.1
.1
.1
.1
522 .25
522 .25
5 1
5 1
R
km
5
1
5
1
J
7
9
7
9
cal cm~
.64x10
.55x10
.64x10
.55x10
-8
-6
-8
-6
min"
2.
8.
4.
9.
75xlO"8
69x1 O"6
53x1 O"8
llxlO"6
Q(R)
°K m
69.4
877
28.5
229
AT
°C
4.6
16.3
3.0
8.4
Note that when the wind speed is increased by a factor of four, the
calculated heat islands decrease by somewhat less than a factor of two, as
I/
•y increases with wind speed. Given the crudity of the model for h(x,y), the
model for Fairbanks is in reasonable agreement with observation. This in
turn suggests that the mean wind through the Fairbanks area is probably
within or below the suggested range of .25 to 1 m sec , with the upper end
of this range, which would correspond to the 1 km radius, being very unlikely.
The case with fog causes problems because of the radiative cooling at
the top. Figure 55 showed a hypothetical fog sounding. The algebraic area
between the original and final soundings at y must equal Q(y) as before.
137
-------
But now (for the constant lapse rate r)
Q(y) . M£_ . <*&™ , or
Q(y) = 2 <54)
The exact relationship between Z and A6 is needed to complete this equation,
but that relationship depends on the details of radiation and convection, as
it will depend on the vertical growth of a fog layer capped by a very steep
inversion. We can, however, set some limits.
Let A81 be the temperature difference and Z1 the mixing depth at y for
the fog-free case. It is obvious from Figure 55 that A6 >_ Ae', as the area
(A61
= Q(y)- Suppose that AS = IDAS' (m not necessarily an integer). Then
2ZmA6'-rZ2
A91 = rZ1
and rZ'2 = 2ZmrZ' - rZ2
Z2 - 2mZZ' + Z'2 = 0
The minus sign corresponds to the case with a superadiabatic lapse rate
above Z, as may be seen from Figure 55 by noting that if Ae = mAe1, mZ1 is
the point at which the foggy sounding and the original sounding cross. So,
if
A6 = mA6',
Z = (m + /m2-!) Z1.
(55)
An upper limit to m may be obtained by noting that Z can hardly exceed
the value it would have if all of the heat that went into the air were
allowed to mix while suppressing radiative loss, and the radiative loss were
then taken all at once. This would give the sounding shown as a dotted line
in Figure 60 as an intermediate step. In this case we can define
138
-------
o>
OJ
X
/
'e-
Figure 60. Development of city fog sounding.
H(y) =
h(y)
pcpv
dy
and proceed in exact analogy to eqs. (12) through (18) to obtain
Z ='
/2H(y)
(56)
(57)
or
m
(58)
(59)
Values of /J?K and m are given in Table 24.
The effect of the lapse rate within the fog is to change our defining
equation for q, eq. (81)
13D
-------
to q = h - 4aTQ3 (A! - .01Z) (60)
TABLE 24. MAXIMUM RATIOS OF FOGGY/CLEAR VALUES OF MIXING DEPTH
AT(m)
K
J
.01
.05
.1
.3
.5
.7
.9
.95
.99
J
'K
10
4.5
3.16
1.83
1.41
1.20
1.05
1.03
1.005
m
5.05
2.35
1.74
1.19
1.06
1.02
1.001
1.0003
1.00001
If a linear relationship exists between AT and Z, this has the effect of
changing the a. In the lower limit case of Z = Z1, for instance, Z = AT/r
and a must be replaced by (1 - ^j-)a. This primarily affects A and thus K/J.
In the extreme case that K/J = 0, this would increase AT by a factor of
The sensitivity of the model to the way a given amount of heating is
distributed through a city is definitely worth noting. For the case with a
linear lapse rate and the "conical" distribution of heat addition discussed
above, it is of some interest to work out the dependence of the model on popu-
lation , wind speed v, potential lapse rate r, energy consumption per person,
s, and the area per person, G = ?rR /. From eq (19) we have at the city cen-
i/
ter AT(r) = /2rQ(R). Consider first the case that j £ 1•
Q(R) = KR2/(2pcpv) = KG(fr/(2pCpvir), (61)
„ i , 3E _ 3s* _ 3siAT
N. - J O- ~ O ~ 1/0 1/9
nO n-J r%J/£-i'/£-
?rK irK u 1/4 G"1/4 s1/2 (63)
provided the units are compatible.
140
-------
-1-1 ? 1
For r in °K m , v in m sec , <|> in persons, G in m person , s
in cal person' min" , p in kg m" and c in cal gm °K , this becomes
AT = 9.2 x IP"3 rl/2 sl/2 ^1/4 G-l/4 y-l/2. (63')
^V K
Likewise, in the radiation-dominated case -y = 0, AT = RJ from (7)
- T
and (39a), giving 4oT
AT = _ . (64)
-1 -1 2 -1
If s is in cal person min and G is in m person , this may be
written as T _~
4.53 x 10"2 (•—} sG"1 (64 ')
I/
For cases with 0< j < 1, the most common state, the situation is
more complex. We have
AT = /2rQ(R) = /rKR2/Pc v = /rKG~1/2) (66)
j u
p
where K = [t + AB + /2AB + (AB, .,-1 .
where the units in (67) are the same as for (631), with the addition
that T is in °K and P is in mb.
For .05 <_ AB/J <_ .25 (the half of the transistion zone nearer convective
dominance)
K 1 (AB}-l/4
J ^ 2.8 VJ '
AT I r3/8 G-7/16 *3/16 v-3/8 s5/8
^ 4/ABAT
141
-------
and AT 2 r1/4 G'5/8 ^/8 v'1/4 s3/4 . (69)
The factor which determines which of the various equations above will be
I/
valid in a particular case is ~, which, from (46) and Table 21 , is a
AD AD
monotonic decreasing function of -p-. ^ = 10 gives a case which is 95%
AB
dominated by radiation; -y- = .001 gives a case 95% dominated by conduction.
AB
As an illustration of the values of ^p to be expected in real cases, Table
25 was made up from Table 4.3, p 58 in the SMIC report (1971). Note that
the values given for s are based solely on anthropogenic heating and may be
serious underestimates in temperate-zone cities where such factors as albedo,
evapotranspiration, and thermal lag of the substrate must also be taken into
account. Given reasonable values of r and v (.1 <_ r v" < .002) most values
I/
of j will fall between about .2 and .8. Thus the transitional case, with
exponents changing, is likely to be the most important one, giving for the
general case
4T . 9.2 x ID'3 r1/2 s1/2 ^4 G'1/4 v-1/2 £ (70)
Table 26 and Figures 61-63 were set up to show the effect of changing
one variable at a time in a hypothetical city with a population of 200,000
2-1 5-1
persons, area per person of 400 m person , heat release of 10 cal person
min"1 (^ 7 KW person"1), potential lapse rate of .03°K~1 (~p = 2°/100m)
-1
and wind speed of 1 m sec . In each case the light line gives the re-
lationship if radiative transfer is ignored; the heavy line includes the
I/
effect of varying T.
-1 3/2 1/2 -1
In summary, if rs G ' v is sufficiently small, AT is propor-
tional to r ' v ' G s ' , in qualitative agreement with previous
-1 3/2 1/2 -1
work. As rs G ' v increases, the influence of r, v and on the
heat island intensity dirnishes and that of G and s increases until in the
extreme case of radiative equilibrium the heat island is due entirely to the
energy release per unit area. At this point, however, the circulation
induced by the heat island would no longer be negligible, and the analysis
is inapplicable.
142
-------
ro
•3-
LU
CQ
«^
t—
^—
J^^
en
O
£
to
o
o
UJ
to
CQ
to
UJ
a:
CO
to
c
o
CO
s_
01
Cl-
lO
o
c
-e-
j_
0
I
C
o
to
S-
CL
c\j
E
CU
r~)
S
•e- o_
0
£=
fV v
>
CO
o
1
^
CM
,__
-o-
cn
, —
to
ro
to
CM
ro
to
to
1 —
X
CU
i a.
o o
4-> O
en
c c
••- o
-C 4J
to co
1X3 O
2 CO
1
P-,
ro
i
>
t-,
CM
f-^
-e-
co
in
•—
CM
O
CM
CO
«>
^f.
O
• —
c
cu
(O
•U
to
cu
c
•r—
S-
T3
i.
O
1
p-
to
1
>
CM
^_
CO
^<
co
CM
ro
CT>
ro
^
,x^
in
fO
cu
s-
et
r—
S-
to
^^
c
"
cu
E
IO
CO
1— 1 r— t— 1 1
1 > 1 1 > >
1— > I_ > > [^ IL,
1 C-. CO C~ I- CM **
> IT) CO *3-~ CM CD IT)
CM «3" CO i—
CM
f-^
f~— r*— 1
i — i — It — 1 > i —
1 1 > 1 > t-, 1
> > C-. > t-. CM >
C-H I-, CM C~ CM ~-^ (-,
CM CM -v. CM ~^. t— CM
^, -- r— ^^ r— -S- -^
i — i — -e- i — -e- in i —
-e- "Q- to -& o i — -e-
ro r-» i— in co o i—
CO "— CM rO
CO CO i — CM Cn i — CM
^" *3" *^" r^x CM O r—
i — i — CM ro ro
co o r*. co CM in ro
CM O ro O o ro ro
«3- in i— -3- r— CM
f— ^~
ro
o o «* co ro r— o
r^ r^ 10 i — CM i —
*3- to ro ^r
to ro r^ in co -^ ro
in ro t— r-*
>^
4-1
c < )
3 ' — «
o s:
O 00
to to C -K
CU CU •!— *
i — « — ( — C to
CU S- rt) -^
en cj^ en cu 4-> c
C C S I- CO 4-> (0
<: •« o 3 10 -a
O JQ •!-> -C S_
tO 00 00 E O*> C -r-
O O O IO O TS 13
^_
1
>
[_
IO
CM
1
>
P-,
CM
^_
-e-
CM
CM
^_
fo
•—
CM
CO
to
o
^t
1—
tn
[ *
o
O-
cu
S-
00
-C
4-)
to
c
10
JD
S-
lO
U-
1
>
[_
§;
CO
10
r^
>
t-.
CM
—
-e-
o
CO
o
in
CM
ro
«—
o
IO
CO
tj-
en
CM
in
<£>
o
^
r—
CO
S- 4->
ia i.
*-• o
co a.
01
J= S-
&- t/)
O -r-
Z J=
to
C J=
5 i1
S- O
*o o
U- CO
"H
o
0
o
II
a.
«,
o
o
0
II
L.
o
"S
J ^
IO
^_
ia
cu
i
CM
-e-
CM
ro"
tn
, —
i
to
CM
1
O
X
^^
CM
CM
II
CO
< -3
CM
e
o
o
s
•^
c
ia
CM
in
ro
•—
<*-
o
IO
t- t= IO
IO CU
*V- !_
? IO
§^_
IO
3
O
"S "
> c
•i- O
s-
cu -a
T3 CU
to
to 10
•r- CQ
£
OL «
l/l
CU
3
t— •
10
L.
3
O
-C
4->
*
[ '
C
CU
g
cu
s-
01
IO
c
4_>
o
c
cu
s-
IO
01
01
t
3
en
M-
00
C
IO
J3.
i.
IO
U.
•K
-------
2000
opoo
100,000
1,000,000 2,000,000
Population
Figure 61. Dependence of city center heat island intensity on population.
Heavy lines include the effect of thermal radiative loss, light
lines give heat island intensity neglecting radiative effects.
Figure 62. (Below) Dependence of city center heat island intensity on
energy release per person (solid lines) and potential lapse rate
(dashed lines). Light and heavy lines as in Figure 61.
Potential Lapse Rate "c/m
.01
e 4
25 100
Energy Release cal/por;,on minute
300
-------
3
Area per Person, m /person
50 100
1000
3200
25 I
Wind Speed, m/sec
10
Figure 63.
Dependence of city center heat island intensity on wind speed
(solid lines) and area per person (dashed lines). Light and
heavy lines as in Figure 61.
145
-------
TABLE 26. VARIATION OF AT (FROM EQUATION 70) WITH POPULATION, AREA PER
PERSON, ENERGY RELEASE PER PERSON, LAPSE RATE AND WIND SPEED.
TEMPERATURE IS 0°C AND AIR PRESSURE IS 1000 mb
a: G =
S =
r =
V =
* =
K
J =
fiT =
b: <(> =
S =
r =
v =
6 =
K
0 =
400m
100
.03°
1 m
2 x
.933
person"
x 10 cal min person"
K m"1
sec"1
103 2 x 104 2 x 105 2 x 106 2 x 107 persons
.884 .804 .679 .507
= 1.3 2.1 3.5 5.2 6.9 °C
2 x
100
.03°
1 m
50
10 persons
x 10 cal min" person"
K m"1
sec"1
100 200 400 800 1600 3200 m2 person'1
.999 .925 .879 .804 .693 .543 .368
AT = 7.2 5.6 4.5 3.5 2.5 1.7 .9 °C
c: 4 =
G =
r =
v =
S =
K =
J
AT
d: 4 =
G =
S =
v =
r =
K
J
AT =
e: <(> =
G =
S =
r =
v =
K
J "
AT =
2 x
400
10 persons
m person"
.03°K m"1
1 m
25
sec"1
50 75 100 150 200 300 xlO3 cal min^person"1
.647 .734 .777 .804 .836 .857 .881
= 1.4 2.2 2.9 3.5 4.4 5.2 6.6 °C
2 x
10 persons
2 1
400m person"
100
1 m
.01
.881
2.2
2 x
400
100
.03
.25
.647
= 5.6
x 10 cal min person"1
sec"1
•03 .06 .1 .15 .2 °K rtf1
.804 .734 .672 .615 .572
3-5 4.5 5.3 5.9 6.4 °C
10 persons
m person"
x 10 cal min" person"1
°K m"1
•51 2 5 10 m sec'1
.734 .804 .857 .906 .933
4-5 3.5 2.6 1.7 1.3 °C
146
-------
REFERENCES
1. Atwater, Marshall A. The radiation budget for polluted layers of the
urban environment. J. Appl. Met, 10, 1971, pp. 205-214.
2. Atwater, M. A. and J. P. Pandolfo. Tundra Environmental Changes Induced
by Urbanization. Climate of the Arctic, ed. G. Weller and S. A. Bowling,
Geophysical Institute, Fairbanks, 1975, pp. 312-315.
3. Benson, C. S. Ice Fog--Low Temperature Air Pollution Defined with
Fairbanks, Alaska as Type Locality. Cold Regions Research and
Engineering Laboratory Research Report 121, 1970.
4. Benson, C. S. and S. A. Bowling. The Sub-arctic Heat Island as Studied
at Fairbanks, Alaska. Climate of the Arctic, ed. G. Weller and S. A.
Bowling, Geophysical Institute, Fairbanks, 1975, pp. 309-311.
5. Bowling, S. A. Radiative Cooling Rates in the Presence of Ice Crystal
Aerosols. Ph.D. Dissertation, University of Alaska, May 1970.
6. Bowling, Sue Ann, Takeshi Ohtake and Carl S. Benson. Winter Pressure
Systems and Ice Fog in Fairbanks, Alaska. J. of Appl. Met., 7, 1968,
pp. 961-968.
7. Childers, Joseph M. and James P. Meckel. Flood of August 1967 at
Fairbanks, Alaska. Hydrologic Investigations Atlas HA-294. U. S.
Geological Survey, 1967.
8. Duckworth, Fowler S. and James S. Sandberg. The Effect of Cities
upon Horizontal and Vertical Temperature Gradients. Bull. Am. Met.
Soc., 35., 1954, pp. 198-207.
147
-------
9. Dwight, H. B. Tables of Integrals and Other Mathematical Data, fourth
edition. MacMillan, New York, 1961.
10. Fahl, Charles B. Internal Atmospheric Gravity Waves at Fairbanks,
Alaska. M. S. Thesis, University of Alaska, Fairbanks, 1969.
11. Haurwitz, Bernhard. Oscillations in a Basin of Cold Air. Atmosphere,
11, 1973, pp. 141-144.
12. Holmgren, B., L. Spears, C. Wilson and C. S. Benson. Acoustic Soundings
of the Fairbanks Temperature Inversions. Climate of the Arctic, ed.
G. Weller and S. A. Bowling, Geophysical Institute, Fairbanks, 1975,
pp. 293-306.
13. Holty, Joseph, G. Air Quality in a Subarctic Community, Fairbanks,
Alaska. Arctic, 26, 1973, pp. 292-302.
14. Landsberg, Helmut. Inadvertent Atmospheric Modifications through
Urbanization. Weather and Climate Modification, ed. W. N. Hess,
Jr. Willey and Sons, New York, 1974, pp. 726-763.
15. Leahey, Douglas M. and James P. Friend. A Model for Predicting the Depth
of the Mixing Layer Over an Urban Heat Island with Applications to New
York City. J. Appl. Met., 10, 1971, pp. 1162-1173.
16. Ludwig, F. Urban Air Temperatures and Their Relation to Extra-Urban
Meteorological Measurements. Papers presented at the Symposium on
Survival Shelter Problems, American Society of Heating, Refrigerating
and Air-Conditioning Engineers. January 19-22, 1970, San Francisco,
California, 1970, pp. 40-45.
17. Myrup, Leonard 0. A Numerical Model of the Urban Heat Island. J. Appl.
Met. 8, 1969, pp. 908-918.
148
-------
18. Nappo, Carmen J., Jr. A Numberical Model of the Urban Heat Island.
Conference on the Urban Environment and Second Conference on Biometeor-
ology, Am. Met. Soc. Philadelphia. Oct. 31-Nov 2, 1972, pp. 1-4.
19. Ohtake, Takeshi, Studies on Ice Fog. Geophysical Institute Report UAG
R-211 University of Alaska. Final Report APO-00449, June 1970.
20. Oke, T. R. City Size and the Urban Heat Island. Conference on the
Urban Environment and Second Conference on Biometeorology. Am. Met. Soc.,
Philadelphia, 1972, pp. 144-146.
21. Robinson, Elmer and Bell, Gordon B. Jr. Low-level Temperature Structure
Under Alaskan Ice Fog Conditions. Bull. Am. Met. Soc., 37, 1956,
pp. 506-513.
22. SMIC Report. Inadvertent Climate Modification. Report on the Study of
Man's Impact on Climate. MIT Press, Cambridge, 1971. 307 p.
23. Summers, Peter W. An Urban Heat Island Model: Its Role in Air Pollution
Problems, with Applications to Montreal. Paper presented at the "First
Canadian Conference on Micrometeorology" in Toronto, 12-14 April, 1965.
24. Wilson, C. R. and C. B. Fahl. Infrasonic Pressure Waves in the Auroral
Zone. Final Report, contract E22-2768 (N) ESSA. Geophysical Institute,
University of Alaska, 1969.
149
-------
,1 RE3OfiT NO
; EPA-cOO/'4-73-•:•;:/
4. TITLE ANDSUBTITc-E
STUDY OF THE SUSARC'fi; ri£,-\7 I'SuAK
FAIRBANKS, ALASKA
7. AUTHOR,3;
S. A. Bow't inc;
C. S. Benson
9. PERFORMING ORGANIZATION i\i,-,,V;E AND ADDRESS
Geophysical Ins"1;';-",, •-
I Fairbanks, AL 9970':
U. ~-iLPCRT DATP
13, PERFORMING ORGANIZATION CODE
13 &ERFORMINC
REPORT NO.
802999
112. SPONSORING AGENCY NAM% /> iN u A DC ROSS ] 13. TYPE OF REPO RT AN D PER IOD CO VERED
Environmental Scien,;,i= R.riaa':xh ..Lcoratory - RTP, NC j pJJ^Q_B/7.4.-4/?R
Office of Research ui^ Dev L
•y r-
'fe.:ts of f,e"' f-heating modified radiative transfer from other
iv'r'ni;,ic.1 v/in'cer insolation virtually eliminated the effects of
.-',".y •jar.pire.ture cycle; snc'v covor and dorniafK' vegetation
'.;-;:rs;ji fill o.i unimportant, end v-:-r\/ low v/ind syaeds minimized
g.-.ness. The observed steady-state hsat island under clear
was croar.c 10°C, with transient values reaching 14°C. This
e co the extremely steep ground inversions known to exist in
\L,r-a ".ntensity correlated well with tha strength of the
0 r,i
-------
United States
Environmental Protection
Agency
Environmental Research Information
Center
Cincinnati OH 45268
Official Business
Penalty for Private Use
$300
Postage and
Fees Paid
United States
Environmental
Protection
Agency
EPA 335
If your address is incorrect, please change on the above label
tear off, and return to the above address
If you do not desire to continue receiving these technical
reports, CHECK HERE , tear off label, and return it to the
above address
EPA-600/4-73-027
------- |