United States
Environmental Protection
Agency
Environmental Monitoring
Systems Laboratory
Las Vegas. NV 89193-3478
Research and Development
EPA/600/S4-88/009 Apr. 1988
&EFA Project Summary
Wet Deposition and Snowpack
Monitoring - Final Project Report
B. C. Hess, J. E. Rocchio, D. J. Chaloud, L. J. Arent, and J. L. Engels
Accurate measurement of snowfall
is critical to the assessment of acidic
deposition trends, particularly in
areas where snow represents 30
percent or more of the annual
precipitation. Such areas include the
intermountain west, where
precipitation amount and, possibly,
total chemical loading appear to be
correlated strongly with elevation
(Svoboda and Olson, 1986). The
Intermountain area is characterized
by complex topography and
meteorology, heavy precipitation, and
extreme cold. A study was
conducted In the spring of 1987 to
evaluate equipment performance in
complex high altitude terrain. The
instruments selected for evaluation
included the Aerochem Metrics
Model 301 wet/dry deposition
collector, the Belfort Model 5-780
weighing rain gage, and the U.S.
Geological Survey (USGS)-designed
46-cm bulk samplers. The first two
are the standard Instruments
employed in the National
Atmospheric Deposition Program
(NADP) and National Trends Network
monitoring systems; the bulk
sampler is used extensively by USGS
in snowfall monitoring studies. All
instruments were installed on a
platform above the expected
snowpack at the University of Denver
High Altitude Laboratory, located
near Mount Evans in Colorado.
Monitoring was conducted between
March 5 and June 1,1987.
Samples were collected on a
weekly and event (i.e., individual
snowstorm) basis. Collected samples
were analyzed for pH, specific
conductance, water equivalent,
selected metal cations (calcium,
magnesium, potassium, and sodium),
ammonium, nitrate, sulfate, and
chloride. All analyses were
performed in laboratory facilities in
Las Vegas, Nevada. Instruments were
evaluated in terms of interinstrument
variability (precision), operational
reliability, and accuracy comparison
to ground-truth data.
Interinstrument variability was
acceptable for most parameters.
Between-event variability was most
pronounced for pH, specific
conductance, nitrate, and sulfate.
Operational reliability was excellent
for all instruments, although frequent
maintenance was required for the
Model 301 sampler to ensure free lid
movement The Model 301 sampler
lacked sufficient volume capacity for
weekly samples or prolonged, heavy
events. Both the bulk sampler and
Model 301 sampler exhibited
acceptable accuracy. Catch
efficiency of the Belfort was less than
1/3 to 1/2 that of either the Model 301
sampler or bulk sampler. The low
catch efficiency was most
pronounced during events of light,
low-moisture-content snow,
indicating a need for effective
windshielding.
This report was submitted in
partial fulfillment of Contract 68-03-
3249 by Lockheed Engineering and
Management Services Company, Inc.
under the sponsorship of the U. S.
Environmental Protection Agency.
This report covers a period from July
1986 to July 1987, and work was
completed as of December 1987.
This Project Summary was
developed by EPA's Environmental
Monitoring Systems Laboratory, Las
Vegas, NV, to announce key findings
-------
of the research project that Is fully
documented in a separate report of
the same title (see Project Report
ordering Information at back).
Introduction
The overall purpose of the project was
to assess the suitability of selected
collection devices to high altitude,
complex terrain situations. This goal was
addressed through the following specific
project objectives:
1. Estimate interinstrument sampling
variability for two colocated
Aerochem Metrics Model 301
wet/dry deposition collectors by
comparing chemistry and water
equivalent for weekly samples.
2. Estimate interinstrument sampling
variability for two colocated Belfort
weighing rain gages by comparing
water equivalent for both weekly and
event samples.
3. Estimate interinstrument sampling
variability for two colocated bulk
samplers by comparing the
chemistry and water equivalent for
weekly samples.
4. Estimate the accuracy of all
collection instruments by comparing
the sample chemistry to the
chemistry of a ground-truth
standard (snow cores). The
comparisons include samples
collected after events and samples
collected weekly.
5. Estimate the accuracy of all
collection instruments by comparing
the water equivalent of samples to
the water equivalent of a ground-
truth standard. The comparisons are
made on snow core measurements,
for both weekly and event samples.
6. Assess operational reliability in
qualitative terms of frequency and
type of instrument malfunctions,
length of downtime, cause and
resolution of problems, ease of
operation, frequency and ease of
maintenance, and evidence of
sample contamination.
7. Recommend use of particular
instruments and sampling intervals
for high altitude, complex terrain
situations based on results of the
above comparisons.
Participants in the study included the
National Acid Deposition Program
(NADP), the U.S. Environmental
Protection Agency (EPA) Region VIII, the
EPA Environmental Monitoring Systems
Laboratory - Las Vegas (EMSL-LV),
and the U. S. Department of Agri-
culture - Forest Service. EPA EMSL-
LV and its prime contractor, Lockheed
Engineering and Management Services
Company, Inc. (Lockheed-EMSCO),
had primary responsibility for
construction of the monitoring platform,
equipment installation, field station
operation, chemical analyses, data
verification and interpretation, and quality
assurance. The monitoring site was
located at the High Altitude Laboratory
operated by the University of Denver.
The High Altitude Laboratory is located
adjacent to the Mount Evans Highway
near Echo Lake, 14 miles south of Idaho
Springs, Colorado. The site offered
several advantages: the terrain is
complex, the area is subject to large
amounts of precipitation and to high
winds, the site is accessible year-round,
it has electrical power, and it has
dormitory facilities. All field operations
were conducted between March 5 and
Junel, 1987.
Procedure
An octagonal platform was
constructed at the site to provide a base
above the expected height of the
snowpack for the following instruments: 3
Aerochem Metrics Model 301 wet/dry
deposition collectors, 2 Belfort Model 5-
780 weighing rain gages, and 2 USGS-
designed 46-cm diameter flanged bulk
samplers. Science Associates Models
424-1 and 424-2 wind speed and wind
direction sensors were mounted on a 2-
m (approximate) mast in the center of the
platform to provide supplementary
meteorological data. The Model 301
samplers and the meteorological sensors
were interfaced to an IBM-PC data
acquisition system (DAS) located in a
heated building approximately 70 meters
from the platform.
Manual ground-level measurements
provided a standard for comparison to
the platform-mounted instrumentation.
Measurements of snow density were
taken in a snowpit. Snowboards provided
a base for collection of snow cores to a
marker horizon on an event and weekly
basis. The snowpit was located in a
separate clearing adjacent to the
platform; snowboards were located in
both clearings.
The monitoring site was serviced by
one full-time operator who resided at
the site in dormitory facilities leased from
the University of Denver. Residence on
site allowed for continuous observation of
weather and instrumentation throughout
the study. Continual observation was vital
to accomplishing the project objective of
assessing instrument reliability and
suitability for snow monitoring in
complex, high altitude terrain. The site
operator was responsible for
instrument checks and calibrate
sample collection and shipment, 1
operation, ground-truth measureme
and documentation of all field activitie
Samples were shipped weekly to
Vegas for processing and analy
Processing included determination
water equivalent, measurement of pH
specific conductance, and preparatio
aliquots for subsequent analy
Analyses, performed every two to
weeks, included measurement of rr
cations (calcium, magnesium, potassi
and sodium), nitrate, sulfate, chlor
and ammonium. A description of
sample aliquots and analytical methoc
presented in Table 1.
Quality assurance (QA) procedi
were integrated into the field, laborat
and data operations. Development of
quality assurance plan began \
development of the project plan
definition of the specific pro]
objectives. Each activity was perfon
in accordance with written protocols
incorporated quality control checks
quality assurance/quality control samp
Field and laboratory data w
compiled in a single data base on
IBM-PC. All data were reviewed
compliance with QA/QC objectives f
to any interpretation of results. Evalua
of QA/QC data was completed as <
were received to identify and rectify
problems. Acceptance criteria w
applied to audit and blank samp
duplicate analyses, and holding times
Verified data were analyzed
accordance with the project object!
Comparisons were made of results
pairs of the sample collection devi
(Belfort rain gages, bulk samplers, Mi
301 samplers, and duplicate snow co
collected over the same sampling pei
The mean, range, percent rela
standard deviation (%RSD), and anal
of variance were computed for w
equivalent and chemistry values.
Comparisons between differ
instrument models employed statist
tests similar to those described above
instruments operating over the s<
sampling interval were compared be
on water equivalent and chemi:
results. In addition, comparisons vt
made of same model and different mi
instruments operating over diffei
sampling intervals.
Results and Discussion
Three of the project objecti
(numbers 1, 2, and 3) were assessrm
of variability related to sampling with
-------
Table 1.
Aliquot Descriptions and Analytical Methodologies
Chemical Parameters
Aliquot Description
Analytical Method
Maximum Holding Time
Project Holding Time
Chemical Parameters
Aliquot Description
Analytical Method
Maximum Holding Time
Project Holding Time
Chemical Parameters
Aliquot Description
Analytical Method
Maximum Holding Time
Project Holding Time
Chemical Parameters
Aliquot Description
Analytical Method
Maximum Holding Time
Project Holding Time
ALIQUOT 1 - CATIONS
calcium, magnesium, potassium, sodium
125-mi Nalgene bottles (acid^washed), filtered (0.45 fim HA type filter),
preserved with ultrapure nitric acid to pH<2
Atomic absorption spectroscopy* (potassium, sodium)
Inductively coupled plasma emission spectroscopy* (calcium, magnesium)
6 monthsb
4 weeks
ALIQUOT2 -NITRATE AND SULFATE
nitrate, sulfate
125-mL Nalgene bottles (not acid-washed), filtered (0.45 fim HA type filter),
preserved with mercuric chloride (0.15 M)
Ion chromatography*
7 days (nitrate)?, 28 days (sulfate)
4 weeks
ALIQUOT 3 - CHLORIDE
chloride
125-mL Nalgene bottles (not acid-washed), filtered (0.45 ftm HA type filter),
no preservative
Ion chromatography3
28 days*
2 weeks
ALIQUOT 4 -AMMONIUM
ammonium
125-mL Nalgene bottles (acid-washed), filtered (0.45 itm HA type filter),
preserved with ultrapure sulfuric acid to pH<2
Flow injection analysis colorimetryd
28 days3
2 weeks
a U.S. EPA, 1983
b Although the EPA (U.S. EPA, 1983) recommends a six-month maximum holding time for these cations, this study
required that all of the cations be determined within 28 days. This was to ensure that significant changes did not
occur and to obtain data in a timely manner.
c Although the EPA (U.S. EPA, 1983) recommends that nitrate in unpreserved samples (not acidified) be determined
within 48 hours of collection, evidence exists that nitrate in mercuric chloride preserved samples is stable for up to
three months (Suarez et. a/., 1 986).
d The method for ammonium determination (U.S. EPA, 1983) was adapted for use on a flow injection analyzer at
EMSL-LV.
same instrument model. Two aspects of
variability were examined in this context:
(1) variability between paired instruments
of the same type and (2) sampling
variability, which is essentially an
examination of the effect of each
instrument model on system precision.
The project objectives specified
comparisons between (1) the Model 301
units used to collect weekly samples; (2)
the bulk samplers, also used to collect
weekly samples; and (3) the Belfort rain
gages used for event and weekly
cumulative water equivalents. In addition
to these specifically stated objectives,
interinstrument sampling variability was
determined for the Model 301 units used
to collect event samples, as was
sampling variability in the collection of
weekly and event core samples.
The sampling variability measured as
percent relative standard deviation
(%RSD) of each parameter per sampler
was determined by pooling the variance
of the sample results for each sampling
date and dividing the pooled standard
deviation by the grand mean. The
residual error term from the analysis of
variance (ANOVA) (Steel and Torrie,
1960) gives the pooled variance, which is
divided by the grand mean to give a
pooled %RSD (coefficient of variation).
These %RSD values, presented in Table
2, represent the relative variability of
each of the sampling methods used.
Snow core data were included in this
-------
Table 2. Interinstrument Sampling Variability.
Instrument Corr.
Type Ca+z Cf Conda H2O EQb
Bulk
9.4
8.56'
9
86.5
0.78
10
3.1
0.60
10
1.4
0.0
11
K*
27.3
0.15
10
Mg*z
11.8
1.12
8
Na*
79.0
1.86
9
NH4"
14.1
0.05
11
N03-
5.0
0.22
11
pH
0.9
0.31
11
S04'2
2.7
0.05
11
Model 301
weekly
21.1
0.65
7
24.5
0.38
6
4.9
0.24
7
8.7
0.93
7
50.6
0.11
6
17.1
0.41
7
15.8
0.03
7
0.0
3.7
1.15
7
1.0
1.87
7
16.6
0.02
7
Model 301
event
18.1
4.16
6
46.4
0.05
8
12.6
0.19
10
10.6
0.08
10
7.9
5.63
6
22.7
3.38
6
14.7
3.27
3
6.2
0.24
4
2.9
0.0
8
4.1
0.39
10
3.6
3.01
8
Be/forts
event
7.5
0.39
10
Be/forts
weekly
5.1
5.54"
11
Core
weekly
31.1
3
94.2
5
72.8
4
7.9
6
100.1
3
54.1
3
58.2
2
16.8
3
14.5
5
2.9
6
45.8
5
Core
event
34.5
4
149
4
86.9
5
13.9
5
24.3
3
47.5
4
9.0
3
33.2
3
33.6
5
2.23
4
15.4
4
Legend
Line 1 gives instrument precision, expressed as percent relative standard deviation (%RSD).
Line 2 gives the "F" distribution for each parameter per instrument type. An asterisk following the value indicates significance at the 95 percen
confidence level.
Line 3 is the total number of sample pairs (n) available for analysis.
"Corr. Cond. = specific conductance, corrected to 25.0°C.
bH2O EQ = water equivalent.
analysis for comparison to the relative
variation of the other methods used.
An additional project objective was to
investigate possible differences between
sampling intervals, specifically by the
comparison of Model 301 event and
weekly sample results. Two of the three
Model 301 samplers were operated on
an event basis over a four-week period;
the remainder of the time, two samplers
were operated weekly. Sampling interval
differences were also investigated for
event and weekly snow core samples.
Event snow cores were collected over
the same four weeks that two Model 301
samples were collected on an event
basis. Snow core data were deleted from
the data base during verification because
the samples they represented were
contaminated or affected by evaporation,
melting, wind scour, or sublimation.
These deletions resulted in an
insufficient number of sample data
remaining to permit an accurate
assessment of variability due to
differences in sampling intervals from
snow cores. Although both event and
weekly data are available for Belfort rain
gages, the weekly data were derived by
cumulation of event data and are,
therefore, not independent.
Water equivalent comparisons were
made by adding individual event data
and comparing the result to the weekly
sample water equivalent. Two events
were collected during the first week of
event sampling. The weekly sample and
the event-cumulative water equivalent
for one of the event samplers were equal
(1.06 cm); the second event-cumulative
was approximately 20 percent less. Only
one event occurred in the second week;
water equivalents for both event
samplers were within 10 percent
agreement with each other and with the
weekly sample. The weekly sample was
the lowest of the three values. A total
seven events were collected in the tr
week. Both of the event-cumulat
water equivalents were greater than t
of the weekly sample; one was about
percent greater while the other was
percent greater. The two eve
cumulatives were within 7 percent
each other. During the last week of ev
sampling, two events were collected. 1
event-cumulatives were equal (2.
cm); the weekly sample was 25 perc
greater. The weekly sample for this I
week, however, included snow from
event which occurred on the day
collection, and the snow from that ev
was not included in the event samples
general, the event-cumulatives were
close agreement with each other and /
the weekly sample.
Instrument reliability as a function
low maintenance and high sam|
recovery was determined from dir
-------
observation of instrument operation by
le field operator and through verification
of sample loss or contamination. The
Model 301 samplers required the most
time for operational maintenance. Over
the first two weeks of sampling, the wet
buckets in each unit were replaced three
times a week to ensure complete sample
collection. During the event sampling
period one particularly heavy snowstorm
required four bucket replacements. The
Model 301 samplers also required snow
removal during snowstorms from the
peaked roofs and from the lid arm joints
to maintain lid mobility. The only
maintenance required for the bulk
samplers was a daily visual check of fill
heights. Maintenance of Belfort rain
gages included daily clock and pen
inspections, replacement of the
antifreeze-oil mixture every 2-4 weeks,
and monthly calibration checks.
Two of the project objectives
(Objectives 4 and 5) were to evaluate the
accuracy of the platform-mounted
instruments by comparing water
equivalent and chemical data to the data
produced by ground-truth measure-
ments. Comparisons were made between
ground-truth measurements and weekly
Model 301, bulk, and Belfort samples,
and event Model 301 and Belfort
samples. The ground-truth measure-
ments were selected to be the standard
for comparison because these method-
ologies have been commonly employed
in classic snowpack studies. A
completely randomized design (CRD)
analysis of variance (Steel and Torrie,
1960) was used to test the differences
between the means of snow core data
and the means of platform-mounted
instruments.
The CRD model used is:
Degrees of Freedom
1
n-1
Source of Variation
Random error
Residual error
where,
n = number of pairs of data collected
The null hypothesis in this case is:
There are no differences between ground-truth
measurements and measurements taken by the
platform-mounted instruments.
F-values were significant at the 95
percent confidence level for all
comparisons of magnesium, potassium,
and chloride. Additionally, event sample
F-values were significant at the 95
percent confidence level for specific
conductance, pH, calcium, and sodium.
The number of significant differences
found between platform-mounted
instruments and ground-truth
measurements does not necessarily
indicate a high degree of inaccuracy in
the platform-mounted instruments. In
this study, the snow accumulations on
snowboards were scraped whenever
volume was insufficient for coring.
Analysis of the scrapings produced little
usable data because of sample
contamination and the high imprecision
of the results that was caused, in part, by
the small number of routine/duplicate
pair data remaining after data verification.
Additionally, weekly snow core and snow
scraping data are suspect due to
observed sample loss by evaporation,
melting, wind scour, and sublimation.
Finally, the snow cores are most directly
comparable to bulk samplers as both are
exposed to both wet and dry deposition.
The cores are less comparable to the
Model 301 samplers since the sampler
lid serves to separate wet and dry
deposition.
Because of the poor results obtained
in this analysis and the suspicion that the
snow core data were questionable, an
additional analysis was conducted. In this
analysis, the results obtained from the
platform-mounted instruments were
compared to each other. This analysis
provides an indication of comparability
among the tested collection methods.
The data indicate that the water
equivalents measured by the Belfort
were consistently lower than those
measured by either the bulk or Model
301 samplers. Belfort sample values
ranged from only slightly lower than
other sample values when wet snow
events were collected to less than 50
percent collection for dry snow events.
This circumstance was probably caused
by air turbulence created by the Belfort
sampler itself. Alter windshields were not
used on the Belfort units because
evidence exists that they are not
effective at windspeeds greater than
3mph (Goodison et al., 1981; Goodison
and Metcalfe, 1982). The low catch
efficiencies observed in this study,
however, indicate that some effective
type of windscreen is needed if Belfort
rain gages are to be used to collect
snowfall.
Comparison of the analytical results
for Model 301 samplers and the results
for bulk samplers indicated the values
were statistically the same for every
parameter measured. The bulk sampler
is exposed to both wet and dry
deposition while the Model 301 sampler
has a lid which prevents dry deposition
from entering the wet bucket. The results
of this analysis indicate that, at least for
this study, dry deposition was not a
significant factor.
Conclusions and
Recommendations
The Belfort rain gage did not perform
well in this study. While interinstrument
precision was acceptable, comparisons to
the bulk and Model 301 samplers as well
as to event snow cores indicated a very
low catch efficiency by the Belforts.
Catch efficiency was lowest during dry
snow events, indicating the problem is
likely due to wind turbulence and,
possibly, could be corrected by addition
of an effective windscreen. Additional
disadvantages noted for the Belforts
included snow accumulation in the funnel
and excessive pen vibration. The first
was resolved by addition of heat tape
which was manually activated during
snow events. The pen vibration is likely
to be a problem any time platforms are
used. Finally, the rain gage produced
only water equivalent data; chemical
analyses were precluded by the use of
an antifreeze-oil mixture in the
collection bucket. For these reasons, the
rain gage as presently configured cannot
be recommended for snow collection in
high altitude, complex terrain.
The coring methodology, while
appropriate to classic snowpack studies,
is not appropriate to short interval
monitoring. Snow accumulation for a
single event or weekly period was often
insufficient for cores. Compositing of
multiple cores or area scrapings as done
in this study increases the contamination
risk. As an alternative, collection of
snowfall from clean plastic sheets placed
on the snow surface might be preferable.
The Model 301 sampler performed
satisfactorily. Both interinstrument and
intermodel precision were acceptable for
most parameters. No samples were lost
due to contamination. The voltage output
of the lid arm mechanism provided data
on the start, end, and duration of events.
The moisture sensors of the three units
were observed to respond individually to
the onset and end of events, but this
variability did not appear to affect the
interinstrument precision significantly.
Modification of the sensor improved
detection sensitivity. A similar sensor is
available from the manufacturer. The
major disadvantages of the Model 301
sampler are: (1) extensive maintenance is
required to ensure operational reliability
and (2) bucket capacity is insufficient for
heavy snow events or prolonged
sampling intervals. Both disadvantages
are likely to be problems in any
-------
unattended operation in heavy snowfall
areas.
The bulk sampler also performed
satisfactorily. The bulk sampler required
no maintenance apart from collection bag
changes, and bag volume was more than
sufficient for weekly sampling.
Additionally, the bulk sampler does not
require electrical power, thus permitting
location in remote areas.
Some disadvantages of the bulk
samplers were noted. The bags supplied
by GSA became brittle at low
temperatures and occasionally leaked;
use of a different material is
recommended, and strength and
temperature testing should be performed
before utilizing bags of a new material.
The bulk sampler is open to the
atmosphere at all times; therefore, it is
subject to dry deposition, and the risk of
contamination is somewhat greater than
for the Model 301 sampler. However,
interinstrument precisions obtained in
this study were nearly equivalent to
those obtained by the Model 301
sampler; comparisons between the bulk
and Model 301 samplers indicated
statistically equivalent measurements of
all parameters. Therefore both the Model
301 sampler and the bulk sampler can
be recommended for snow collection.
References
Goodison, B. E., H. L. Ferguson, and G.
A. McKay, 1981. Measurement and data
analysis. Handbook of Snow - Prin-
ciples, Processes, Management, and
Use, Edited by Gray, D. M. and D. H.
Male. Pergamon Press, Willowdale,
Ontario, pp. 191-274.
Goodison, B. E., and J. R. Metcalfe,
1982. Canadian Snow Gauge Experiment
Recent Results. Proceedings of the
Western Snow Conference, Reno,
Nevada. April 20-23, 1982.
Steel, R. G. D., and J. H. Torrie, 1960.
Principles and Procedures of Statistics.
McGraw-Hill, New York.
Suarez, F. X., D. C. Hillman, and E. M.
Heithmar, 1986. Stability of nitrate in
preserved and unpreserved natural
surface waters. Presented at the Rocky
Mountain Conference on Analytical
Chemistry, August 3-5, Denver,
Colorado.
Svoboda, L. and R. Olson, 1986. Quality
Assurance Project Plan for the Rocky
Mountain Deposition Monitoring Project
as Part of the Western Conifer Research
Cooperative. U.S. Environmental Pro-
tection Agency, Environmental Re-
search Laboratory, Corvallis, OR,
unpublished results.
U. S. Environmental Protection Agency,
1983 (revised). Methods for Chemical
Analysis of Water and Wastes. EPA-
600/4-79/020. U.S. Environmental Pro-
tection Agency, Cincinnati, Ohio.
. S. GOVERNMENT PRINTINGOffKE: 1988/548-158/67105
-------
— w
fi.C. Hess, J.E. Rocchio, D.J. Chaloud, L.J. Arent, and J.L Engels are with
Lockheed Engineering and Management Services Company, Inc., Las
Vegas, NV 89119.
W.L Kinney is the EPA Project Officer (see below).
The complete report, entitled "Wet Deposition and Snowpack Monitoring - Final
Project Report," (Order No. PB 88-165 717IAS; Cost: $19.95, subject to
change) will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Environmental Monitoring Systems Laboratory
U.S. Environmental Protection Agency
Las Vegas, NV 89193-3478
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
Official Business
Penalty for Private Use $300
EPA/600/S4-88/009
0008329 f>$
------- |