Environ. Sci. Technol. 1988, 20, 275-290
Snow Chemistry of the Cascade-Sierra Nevada Mountains Leslie 6. Laird,” Howard E. Taylor,? and Vance C. Kennedy* US. Geological Survey, Water Resources Division, Pacific Northwest District, 1201 Pacific Avenue-Suite Washington 98402
This investigation assesses geographic variations in atmospheric deposition in Washington, Oregon, and California using snow cores from the Cascade-Sierra Nevada Mountains, collected from late February to mid-March 1983. A statistical analysis of the analytical and sampling precision was made. The snowpack in the higher Cascades and Sierra Nevada is not strongly influenced by anthropogenic activities at present. The pH of snow samples ranges from 5.11 to 5.88. Sulfate and nitrate correlate with H+ in some segments of the sample traverse. The SO4data show apparent influence from major source areas in Washington and California; nitrate does not. An apparent decrease in NH4 in snow in Washington and California suggests atmospheric interactions resulting in the removal of NHI. The NH4 reduction raises questions about nutrient supply to the mountain vegetation. Heavy-metal correlationsincluded Cd, Cu, and Fe with Pb, and Mn with K and DOC, among others. No correlation was found between constituents and snow-water content.
600, Tacoma,
A total of 102 snow cores were collected 70 in the main north-south transect, duplicates at 16 of the north-south sampling sites, 10 samples in east-west transect lines, and 6 samples of surface snow. The results presented in this report are based primarily on the 70 samples collected in the north-south transect. This report deals only with the solutes that were found in the snow. The insoluble particulate matter in each sample was carefully separated and will be the subject of future examinations. The constituents and properties reported here are pH, hydrogen ion (H+),calcium (Ca), magnesium (Mg), sodium (Na), potassium (K), chloride (Cl), sulfate (SO,), nitrate (NO,), fluoride (F), phosphate (PO,), ammonium (NH,), iron (Fe), aluminum (Al), manganese (Mn), copper (Cu), cadmium (Cd), lead (Pb), dissolved organic carbon (DOC), and specific conductance. Analyses were also made for bromide and arsenic, but concentrations were below the detection limit (Table I). Methods and Quality Assurance
Introduction
The chemistry of atmospheric precipitation along the West Coast of the United States has become of increasing interest in recent years. Before the 19709, it was generally assumed that the prevailing westerly winds from the Pacific Ocean carried no significant pollution from upwind sources and that acid deposition, or “acid rain”, and related problems were localized and not prevalent over large regions. However, recent data collection and research have shown increasingly a pattern of deleterious atmospheric deposition in the Pacific Coast States, although most of these studies have dealt with rather limited geographic areas (1-5). Our investigation was designed to assess geographic variations and trends in atmospheric deposition in Washington, Oregon, and California by sampling the snowpack in the Cascade-Sierra Nevada Mountains from the U.S.-Canadian border to a point northeast of Bakersfield, CA (Figure 1). The study was prompted by concern about the potential for acidification of sensitive lakes and streams in the Cascade-Sierra Nevada Mountains and associated impacts on the forests of this region, as well as by the need to further identify the depositional pattern of constituents from natural and anthropogenic sources. Snow sampling was selected for several reasons: the snowpack provides an integration of precipitation events under what are probably varying conditions of natural and anthropogenic water-soluble and particulate-chemical inputs; the snowpack is an accumulator of dry fallout during nonstorm periods; the snow accumulates during the wet winter season and contains much less wind-blown dust and other terrestrial material than at other times of the year (5); and a large geographic area could be sampled and analyzed more readily by this approach. Present address: U.S. Geological Survey, Denver Federal Center, MS 407, Lakewood, CO 80225. Present address: U.S. Geological Survey, MS 496, 345 Middlefield Road, Menlo Park, CA 94025.
Sampling of the mountain snowpack near maximum accumulation, but before spring melt started, was deemed to be the simplest and most effective method of assessing wintertime depositional patterns of airborne chemicals in this region. However, use of snow data has some shortcomings. The Cascade and Sierra Nevada Mountains have a “warm” snowpack with temperatures usually close to 0 “C (6). Because of air-temperature variations, some melting of the snowpack often occurs during the winter, and depending on temperature conditions, rain may fall on the snowpack and percolate through it. Such percolation, if it continues through the entire depth of the snowpack, can leach soluble material from the snow in concentrations disproportionate to those in the snowpack (7,8). In addition, Colbeck (7) found that the rate at which impurities are removed from the snow depends upon the atmospheric conditions under which the snow was deposited, as well as the degree and type of metamorphism the snow has undergone and the intensity of rain and f or melt events. Thus, the snowpack cannot be assumed to accumulate and hold all atmospheric deposition during the life of the snowpack, nor can it be assumed to lose chemicals due to surface melting or rain on the snow. However, to lessen the possibility of rain and melting impacts, almost all our sampling sites (66 of 70) were located above 1680 m in elevation,and sampling occurred in late winter before the spring melt of the snow began. Wherever possible, the sampling sites were located in the southwestern part of meadows or in open areas where shading occurred to lessen the effect of surface melting from solar radiation. In addition, to assess the loss of chemicals from the snowpack, four 2 X 2 m waterproof boxes were set out in November, prior to major snowfd, and snow cores were collected from the snow column inside and outside of the boxes during the late winter sampling period. No significant variation in snow chemistry was found between cores in each set (data presented later in this report). For these reasons, the snow-core data for the entire transect are believed to allow reasonable comparisons of variations and geographic trends in soluble substances in the atmospheric deposition.
Not subject to US. Copyright. Published 1986 by the American Chemical Society
Environ. Sci. Technol., Vol. 20,
No. 3, 1986 275
.
Washington
X.
.*
.:
Oregon
Figure 1. Snow sample sites.
Snow cores were taken from late February to mid-March 1983 at approximately 25-km intervals just west of the general crest of the mountain ranges, from the US.-Canadian border to a point in southcentral California. Sampling was done at the highest elevation practicable in the individual areas. Sampling elevations ranged from 1280 to 3140 m; the median was 2040 m. The snow cores were collected by using a stainless-steelsampler of Federal Snow Sampler design (cutter head opening 3.77 cm). The lower 7-8 cm of the sample was routinely discarded to eliminate contamination from the underlying soil or rock. Several complete cores, taken to a polyethylene surface at the base of the snowpack, included the lower 7-8 cm of snow and were analyzed for the chemical constituents and properties listed above. These were compared to adjacent cores from which several centimeters of snow at the bottom were discarded. There were no significant differences. All samples were double bagged in polypropylene, heat sealed, and kept frozen by mechanical refrigeration from the point of collection until they were analyzed in the US. Geological Survey laboratories in Denver, CO. Helicopter transport was utilized between all sample sites, and great care was taken to assure that the jet engine exhaust did not contaminate the sampling sites. Sample Preparation and Analytical Methodology. To ensure uniform handling, the melting, filtration, and bottling process followed a strict protocol (9). The snow sample was carefully melted so the overall temperature of the sample never exceeded 4 O C , hence any bacterial deterioration of the nutrient constituents was minimized, as was adsorption of trace metals on the surfaces of the polypropylene bag. Immediately after thawing, the samples were filtered, preserved, and bottled. Samples for dissolved organic carbon were filtered through a 0.45-pm silver membrane filter housed in a stainless-steel holder. All other determinations were performed on samples filtered through 0.4-pm-pore-diameter polyethylene membrane filters. One liter of the filtered sample was collected in a polyethylene bottle, previously rinsed with nitric acid, and reserved for metals analysis. This sample was preserved by the addition of 2 mL of double distilled, high276
Environ. Sci. Technol., Vol. 20, No. 3, 1986
purity concentrated nitric acid. Up to an additional 1 L of filtrate was collected in a deionized, water-rinsed polyethylene bottle for nutrient analysis and was quickfrozen to avoid the addition of preservatives and inhibit adsorption on the container walls. A pH and specific conductance were determined immediately on the filtrate. A variety of methods were used for the analysis of specific constituents in the melted snow cores (Table I). In all cases, standard stock solutions of calibration standards were prepared from ultra-high-purity materials dissolved in distilled-deionized water (acidified with nitric acid) and stored in ultra-high-purity, nitric acid-rinsed (for metals) bottles made of Teflon. Graphite furnace atomic absorption spectrometry technique included five successive 50-pL aliquots of sample dispensed into the graphite furnace tube, with solvent evaporation at 110 OC occurring between each aliquot deposition. This permitted larger quantities of analyte to be present in the tube at the time of atomization, greatly improving precision and sensitivity. Ammonium and NO3 ions were determined colorimetrically, immediately after the samples were thawed, by specially optimized, low-concentration-level automated absorption spectrophotometry. A low-ionic-strengthglass electrode was used to measure pH potentiometrically. The pH meter was calibrated periodically using special lowionic-strength solutions prepared by diluting a sulfuric acid stock solution. All pH determinations were made concurrently with the specific conductance measurements. Because the samples were extremely dilute, great care was exercised in handling and analysis. To minimize the possibility of systematic errors (i.e., day-to-day variations in instrumentation, standardization, and any other nonrandom sources), no samples were analyzed (except for pH and conductivity) until all preparation was completed. Analyses were then performed as simultaneously as possible for a given parameter. For all constituents, samples were analyzed in a random order, with reference standards and blanks interspersed at a 10% frequency. Duplicate samples, chosen at random and composing 25% of the total sample set, were randomly interspersed with the other samples during the analysis process. The standard reference materials consisted of US. Geological Survey Standard Reference Water Samples and a National Bureau of Standards Standard Reference Material (SRM) sample. As can be seen from Table 11, a comparison of the mean concentration values and the standard errors of the means shows that all parameters determined were well within the experimental error of the analytical procedures. Because each of the samples in this study was collected in a systematic fashion (i.e., the sampling strategy was based on geographic considerations)rather than randomly, it cannot be anticipated that the distribution of concentrations of any specific constituent or a subsample of the distribution is Gaussian in nature. Therefore, the paired sampling technique as described by Koch and Link (19) was employed to establish a normal distribution of test data that allows basic statistical calculations to be performed. Approximately 20% of the prepared laboratory samples, selected at random, were split, and duplicates were renumbered and randomly interspersed with the original sample population. The paired analyses of the duplicates for each constituent were ordered with corresponding samples adjacent to each other. The algebraic differences between the paired values (original minus duplicate) were used to establish a data set that was random in nature and hence followed a normal sample distribution. This was verified by plotting the frequency distribution of these differences of paired values on normal probability
Table I. Methods of Analysis, Detection Limit of Chemical Constituents, and Properties of .Snow Samples detection limit, mg/L
method of analysis (ref)
parameter specific conductance PH Ca Mg Na K
conductance bridgeldip cell potentiometric/glass electrode ICP atomic emission spectrometry (10) ICP atomic emission spectrometry (10) ICP atomic emission spectrometry (10) atomic absorption spectrometry (11) ion-exchange chromatography (12) ion-exchange chromatography (12) fluorescence spectrometry (13) ion-exchange chromatography (12) absorption spectrometry (14) absorption spectrometry (15) absorption spectrometry (14) graphite furnace atomic absorption spectrometry atomic absorption spectrometry (17) graphite furnace atomic absorption spectrometry graphite furnace atomic absorption spectrometry ICP atomic emission spectrometry (10) graphite furnace atomic absorption spectrometry ICP atomic emission spectrometry (10) infrared adsorption spectrometry (18)
so4
c1 Br F N as NO3 N as NH4 P as PO4 A1 As Cd cu Fe Pb Mn DOC
-
0.002 0.003 0.01 0.01 0.01 0.01
0.005 0.01
0.002 0.003 0.002 2 x 10-5 0.001 2 x 104
(16) (16) (16)
5 x 104
1 x 10-4 2 x 10-4 1 x 10-4 0.1
(16)
Table 11. Statistical Data for Analysis of Standard Reference Materials experimental parameter
srmn
srm mean
Ca' MgC Nac
74 74 74 80 80 74 74 N7 69 75 73 1643a 69 1643a
7.48 1.94 2.77 0.437 1.43 0.140 2.43 0.029 3.10 0.079 0.626 88 0.023 31
K' SO4'
C1' N as NO,' P as PO4' Ald Cdd Cud Fed Pbd Mnd
srm SEb 0.15
0.04 0.08 0.008 0.07 0.03 0.11
0.002 0.32 0.004 0.023 4 0.006 2
mean
SE
7.46 1.94 2.81 0.426 1.43 0.141 2.48 0.030 3.37 0.084 0.613 85 0.027 31
0.01 0.01 . 0.01
0.003 0.01
0.007 0.0 0.002 0.08 0.006 0.049 2 0.009 1
Standard reference material. Standard error at 0.95 confidence level (Cu, Cd, Pb, K, Al; SE obtained by dividing SE by dilution factor). 'In mg/L. udL. Table 111. Precision of Analytical Determinations Made on Snow Samples (95% Confidence Interval)
Table IV. Confidence Intervals of Duplicate Samples Based on Overall Precision
parameter
concn"
precision
parameter
concn"
specific conductanceb PH Ca' Mg' Na' K'
2.90 5.56 0.115 0.081 0.293 0.124
10.08
specific conductanceb PH Ca' MgC Na' K'
SO4'
0.017
C1' NOS' N as NH4' Cdd Cud Fed Pbd Mnd
0.258 0.0257
2.65 5.52 0.031 0.018 0.049 0.018 0.14 0.33 0.037 0.065 0.003 0.025 0.103
fO.O1
0.807 0.95 2.68
f0.02 f0.03 fO.01 f0.02 10.02 f0.05 f0.002 f0.008 f0.004 10.2 f0.02
0.400
ai0.02
0.0415
SO4'
1.47
f0.05 " Average concentration of split sample subset. In fiS/cm. mg/L. gg/L.
In
paper. The linear nature of the resulting plot confirmed the hypothesis that the distribution was, in fact, Gaussian and hence represented random behavior. A hypothesis was formulated that if the analytical pro-
CF F' N as NO3' N as NH4' P as PO4' Cdd Cud Fed Pbd Mnd
0.50 1.2
0.291 1.4
precision f0.22 fO.11
f0.02 10.01 10.01 fO.01 f0.03 fO.09 aio.01
f0.004 fO.O1
f0.002 f0.02 f0.2 f0.2 f0.08 f0.5
"Average parameter concentration of duplicate snow cores. In fitS/cm. cIn ma/L. NEIL.
cedure for a given constituent was producing data that were systematically accurate over the entire data set, then the differences in values between replicated samples should Environ. Scl. Technol., Vol. 20, No. 3, 1986
277
Table V. Calcium and Sulfate Concentrations for Box Sites
location/elevation Mount Rainier/l707 m inside box outside box Crater Lake/1966 m inside box outside box Mount Shasta/1768 m inside box outside box Donner Pass/2134 m inside box outside box
concn, mg/L calcium sulfate 0.024 0.011
0.22 0.12
0.013 0.020
0.07 0.10
0.022 0.013
0.31 0.49
0.022 0.013
0.25 0.10
vary only to the extent of experimental error of the methodology and should differ in a random nature. Therefore, the mean of the algebraic differences of the paired determinations should be equivalent to zero. Student's t tests can be used to test for this condition (20). Computation of data sets for each constituent showed that at the 99% confidence level the mean value of the differences of the paired replicates was not significantly different from zero. Therefore, the original hypothesis was not disproven, and it is assumed with a high degree of confidence that all paired analytical replicates for each constituent data set represent the same distribution. This, in turn, suggests that the analytical results within a given analysis set (samples and standards) are fully within the accuracy and precision control established by the nature of the methodology and that no systematic errors are biasing the data. From this set of paired data, precision of analysis was estimated for each constituent. Table I11 lists a calculated precision about an average concentration level in the subset of duplicate samples from the original population. The precision was calculated based upon estimating the average standard deviation from the average range of the paired replicates (21). From the average standard deviation, the average standard error of the mean was computed. From this, the 95% confidence interval was established, which is listed in Table 111. Duplicate samples were collected at 16 of the 70 snow sampling locations. Data for these pairs of samples were analyzed for precision by use of a method based on the technique of Koch and-Ljnk (19), and the precision was calculated based upon estimating the standard deviation of the average range of the paired replicates (21) (as described under the previous section on quality control). Table IV lists calculated overall precisions of field duplicate samples at average concentration levels for each of the constituents in the subset of duplicate samples. These precisions necessarily include the laboratory analysis precisions. Numerous investigators (7,8,22) have reported on the process of the loss of solutes from a snowpack during the winter, particularly during the spring melt period. This sampling program was designed to occur before the spring melt started; but to assess the loss of solutes from the snowpack during the winter, four wooden containers, about 2 m square and lined with plastic, were placed in early winter at locations above an elevation of 1700 m near the north-south sampling route (Figure 1). Immediately after the main body of samples was collected, the boxes were visited and snow cores collected from the snowpack over the box and outside of the box. The boxes were then dug out, and the condition of the snow inside and outside of 278
Environ. Sci. Technol., Vol. 20, No. 3, 1986
Table VI. Statistical Data on Snow Samples
constituent
max
min
specific conductance" PH H+ (mequiv) Cab Mgb Nab Kb
6.10 5.88 0.008 0.16 0.051 0.26 0.30 0.32 1.0 0.10 0.12 0.18 0.022 100. 0.96 3.8 8.5
1.78 5.11 0.0005 0.005 0.003 0.01 0.01 0.04 0.01 0.01 0.002 0.003
Clb Fb N as NO? N as NH,b P as P04b AIb Cd' Cut FeC Pbc Mnc DOCb "In &/cm2.
1.1
7.8 4.1
0.002
0.18 0.024 0.100 0.1 0.05 0.01 0.20
mean
median
2.8
2.52 5.58 0.003 0.36 0.003 0.06 0.01 0.13 0.16 0.03 0.022 0.010 0.004 1.7 0.073 0.28 0.1 0.28 0.7
-
0.003 0.042 0.009 0.07 0.019 0.14 0.22 0.03 0.025 0.08 0.005 3.2 0.12 0.44 0.92 0.33 1.0 0.60
0.50
mg/L. cIn wg/L.
the box examined. In each case, 18 of the 20 constituents and properties analyzed in each pair of samples compared very closely. However, calcium and sulfate results were inconsistent (Table V). Calcium concentration was greater inside the box at three sites and less at one, and sulfate concentration was greater inside the box at two and less at two. The snow at three of the boxes was dry and fine grained with no sign of melting inside or out, but the fourth box (Mount Shasta) contained about 5 cm of slush. The slush in the box may have been produced by melting of the snowpack and/or rain on the snowpack or may have been due to precipitation at the time the box was emplaced. Heavy precipitation, which was a mixture of snow and rain, fell throughout the day that the site was established, and the temperature was close to 0 OC. Outside of the Mount Shasta box, the snow at the soil was fine grained and showed no appearance of metamorphic change, suggesting that the snow near the ground may not have been subjected to melting or percolation of water through the snowpack. Although it is likely that some loss of solutes from the snowpack does occur in many areas during the winter months, we believe the losses from the sample sites selected were not significant for the purposes of this investigation. Results and Discussion
Figures 2-11 show variations in concentration of 18 constituents and properties determined for sites in the sampling line from the US.-Canadian border southward. Only those parameters are shown for which a trend can be observed visually. Table VI presents summary statistics for 20 parameters and properties analyzed for the 70 points sampled during the survey. One of the purposes of this study was to detect trends in composition of the high-mountain snows in relation to probable solute sources. This necessarily implies the influence of meteorologic processes. A second purpose was to understand the relations among the various properties and constituents. These two subjects are discussed below. Meteorological Processes. Wind patterns are important in determining the probable sources of solutes in the snow. The U.S. Pacific Coast has prevailing westerly winds. During the winter, the Pacific high-pressure cell, which dominates the low-levelairflow, weakens and moves south, resulting in weaker and less persistent winds along
.
'1 ble VII. Significance Values Equal to or Exceeding 0.95 for Linear Correlations of Snow Sample Parameters
I * \
EXPLANATION
I I
* I
* II
*
Table Value
* I \ -2
*
II *
1
*
K
* *
*
1
2
*
so4 7
*
*
*
2
* F
*
* NO?
*
NH4
*
- 1
PO^ A1 *
Cd
*
*
*
*
*
*
*
1
*
= =
*
*
*
3
*
* *
*
* *
I
*
*
*
*
*
* Mn 3
* *
*
*
*
*
* *
*
*
* -
*
2
*
*
*
*
*
*
*
*
1
*
I **
*
*
*
*
*
*
*
*
*
*
*
*
*
*
* *
*
*
3
* * *
*
*
*
*
*
*
*
* 3
*
*
1
*
*
*
*
*
* * 2
*
* *
*
*
* 1
*
*
*
*
*
* *
*
*
*
*
*
2
*
*
*
2
*
*
*
* *
*
*
* *
1
*
*
*
*
*
*
*
1
*
*
*
*
**
*
*
*
-1
*
*
*
*
*
*
I**
*
*
*
\
*
****
~
:2**
* *
*
*
*
*
I * * I ****j1ffjI 3
*
*
* -3
1
*
*
*
*
1
*
2
3 - 3 1 3
*
*
*
*
*
