seasonal change in precipitation, snowpack, snowmelt, soil water and streamwater chemistry, northern...

Seasonal change in precipitation, snowpack, snowmelt,soil water and streamwater chemistry,

northern Michigan

Robert Stottlemyer1* and David Toczydlowski21US Geological Survey, 240 W. Prospect Rd, Ft. Collins, CO 80526

2Department of Biological Sciences, Michigan Technological University, Houghton, MI 49931

Abstract:We have studied weekly precipitation, snowpack, snowmelt, soil water and streamwater chemistry throughoutwinter for over a decade in a small (176 ha) northern Michigan watershed with high snowfall and vegetated by60 to 80 year-old northern hardwoods. In this paper, we examine physical, chemical, and biological processesresponsible for observed seasonal change in streamwater chemistry based upon intensive study during winter

1996±1997. The objective was to de®ne the contributions made to winter and spring streamwater chemicalconcentration and ¯ux by processes as snowmelt, over-winter forest ¯oor and surface soil mineralization,immobilization, and exchange, and subsurface ¯owpath. The forest ¯oor and soils were unfrozen beneath the

snowpack which permitted most snowmelt to enter. Over-winter soil mineralization and other biologicalprocesses maintain shallow subsurface ion and dissolved organic carbon (DOC) reservoirs. Small, but steady,snowmelt throughout winter removed readily mobilized soil NOÿ3 which resulted in high over-winter stream-

water concentrations but little ¯ux. Winter soil water levels and ¯owpaths were generally deep which increasedsoil water and streamwater base cation (CB), HCOÿ3 , and Si concentrations. Spring snowmelt increased soilwater levels and removal of ions and DOC from the biologically active forest ¯oor and shallow soils. Thesnowpack solute content was a minor component in determining streamwater ion concentration or ¯ux during

and following peak snowmelt. Exchangeable ions, weakly adsorbed anions, and DOC in the forest ¯oor andsurface soils dominated the chemical concentration and ¯ux in soil water and streamwater. Following peaksnowmelt, soil microbial immobilization and rapidly increased plant uptake of limiting nutrients removed

nearly all available nitrogen from soil water and streamwater. During the growing season high evapo-transpiration increased subsurface ¯owpath depth which in turn removed weathering products, especially CB ,HCOÿ3 , and Si, from deeper soils. Soil water was a major component in the hydrologic and chemical budgets.

Copyright # 1999 John Wiley & Sons, Ltd.

KEY WORDS snowpack; snowmelt; soil water; biogeochemistry; stream chemistry; Michigan

INTRODUCTION

The Lake Superior Basin receives moderate atmospheric inputs of H�, NH�4 , NOÿ3 , and SO2ÿ4 (NADP

1982±1996). Snowfall can equal up to half the annual precipitation (Eichenlaub et al., 1990), and thesnowpack may temporarily store up to 50% of annual precipitation (Stottlemyer 1987). Snowmelt can causesharp changes in streamwater chemistry by dilution and the addition of solutes (Johannessen and Henriksen,1978). In the Upper Great Lakes region, sensitive aquatic systems occur where substrates are dominated byPrecambrian granites and glacial outwash plains (Je�ries et al., 1979; Kelso et al., 1986). But most of the

CCC 0885±6087/99/142215±17$17�50 Received 27 April 1998Copyright # 1999 John Wiley & Sons, Ltd. Revised 24 October 1998

Accepted 18 March 1999

HYDROLOGICAL PROCESSESHydrol. Process. 13, 2215±2231 (1999)

*Correspondence to: Dr R. Stottlemyer, US Geological Survey, 240 W. Prospect Rd, Ft. Collins, CO 80526, USA.

region has soils with high base saturation and surface water with high acid neutralization capacity (ANC)(Wiener and Eilers, 1987).

In snow-dominated ecosystems, establishing the cause of seasonal change in streamwater chemistry is notsimple. Relations between precipitation, snowpack, snowmelt, and streamwater chemistry are subject tomuch temporal and spatial variation. The lateral movement of solutes within the snowpack complicatesextrapolation of results from snowmelt lysimeters (Kattelmann, 1989; Harrington and Bales, 1998). A rapidhydrologic response with snowmelt in headwater streams may not re¯ect only surface runo�. Streamwatercan include deeper subsurface water displaced by snowmelt (Caine, 1989). In high snowfall regions, a majorsource of variation is the degree to which snowmelt enters the soil. Frozen soils delay snowmelt and increasethe amount of snowmelt solutes directly reaching the stream (Pierson and Taylor, 1985). Unfrozen forest¯oors permit snowmelt to enter where its chemistry is rapidly altered (Hazlett et al., 1992). Upon entering theforest ¯oor and surface soil, the alteration of snowmelt or rain chemistry depends upon ¯owpath andresidence time before reaching the stream (Rice and Bricker, 1995).

This paper examines physical, chemical, and biological causes of seasonal change in streamwaterchemistry prior to, during, and following snowmelt at the Calumet watershed, Michigan, for the winter of1996±1997. The objective was to partition out the roles processes such as snowmelt, over-winter forest ¯oorand surface soil mineralization and immobilization, forest ¯oor and soil ion exchange, and subsurface¯owpath have in regulating winter and spring streamwater chemical concentration and ¯ux. This short-termstudy was intended to address questions raised from ongoing, longer term watershed research (Stottlemyerand Toczydlowski, 1996a, b).

METHODS AND MATERIALS

Site description

The Calumet watershed (176 ha) is a ®rst-order catchment vegetated by 60±80 yr old sugar maple(Acer saccharum) and white birch (Betula papyrifera) (Figure 1). The watershed is adjacent to Lake Superior,has a northwest aspect, uniform slope, and 195 m relief. The bedrock is predominantly Cambrian Fredasandstones overlain with alkaline till and old beach deposits. Soils are primarily Typic Haplorthods, sandy,mixed, frigid. Dominant horizons are O (7±0 cm), A (0±5 cm), E (5±13 cm), Bhs and Bsl (13±62 cm),BC (62±98 cm), and C (98 to4200 cm). An almost impervious soil layer (ortstein) occurs at 1.5±2 m depth.Weathering of the alkaline till results in moderately bu�ered streamwater (595 meq Lÿ1) (Stottlemyer andToczydlowski, 1991). Climate on lake Superior's south shore is modi®ed by the lake which rarely freezesover. Open water moderates winter temperature extremes and contributes to the region's high (4500 cm yrÿ1)snowfall. These conditions, coupled with daily maximum winter air temperatures approaching 0 8C andunfrozen soils, result in periodic midwinter thaws and small, but steady, snowmelt throughout much ofwinter. The dominant streamwater ions are Ca2� and HCOÿ3 , and in precipitation H� and SO2ÿ

4

Field methods

Field and laboratory methods are summarized here and detailed in Stottlemyer and Toczydlowski (1991,1996a, b). Four precipitation and snowpack monitoring stations are located at C4 (190 m) near the elevationof Lake Superior, C3 (240 m), C2 (295 m) and C1 (350 m) (Figure 1). In winter each station is equipped witha bulk precipitation collector. In summer, precipitation is collected at station C1 using a bulk collector. Drydeposition is not measured. A snow survey transect (®ve sampling points) is set up at each station wheresnow water equivalent (SWE) and depth are determined using a Mount Rose snow sampler. At each stationa single snow core is collected for chemical analyses using a 5 cm diameter PVC tube. Stream dischargeis measured year round at the mouth of the watershed using a 30 cm wide Parshall ¯ume equipped withpressure transducer and datalogger. Stream discharge is also recorded at headwater station C21 in winter andspring.

Copyright # 1999 John Wiley & Sons, Ltd. HYDROLOGICAL PROCESSES, VOL. 13, 2215±2231 (1999)

2216 R. STOTTLEMYER AND D. TOCZYDLOWSKI

In 1987, three 2� 3 m snowmelt lysimeters were installed near station C5. Snowmelt is recorded byunderground tipping bucket raingauges. Near each snow lysimeter, triplicate tension soil lysimeters wereinstalled at depths of 15 and 30 cm in mineral soil. Soil water sampling wells were installed to the depth of theortstein layer at stations C5 (200 cm) and C21 (190 cm). Since the plots were located in the bottom half of thewatershed, the snowpack and snowmelt season is somewhat shorter than higher elevation stations.

Samples for chemical analyses were collected weekly. During snowmelt, stream samples from justupstream of the ¯umes were collected daily at maximum and minimum discharge, and snow lysimeters weresampled several times each week. The wells and tension lysimeters were sample weekly or more frequentlyduring rapid snowmelt. The wells were pumped out and allowed to re®ll before collecting a sample. Duringthe snow-free season, streamwater and precipitation were sampled weekly.

A National Atmospheric Deposition Program (NADP) site, 21 km from the watershed, has been inoperation since 1982. The National Oceanographic and Atmospheric Administration (NOAA) operates aprimary weather station at the Houghton County Airport 6 km from station C1 and at the same elevation.These additional stations provide reference data for variables monitored at Calumet.

Figure 1. Station locations in Calumet watershed, Lake Superior Basin, Michigan


SNOW HYDROLOGY 33: SEASONAL CHANGES 2217

Laboratory methods

Samples were immediately transported to the laboratory at Michigan Technological University. Snowsamples were melted at room temperature in pre-rinsed, covered polyethylene containers. PH, speci®cconductance, and alkalinity (titration with 0.02 N H2SO4 to pH 4.5) were determined when samples reachedroom temperature (524 hours). Filtered (pre-rinsed 0.45-mm) subsamples were refrigerated (2 8C) for up tothree weeks. Concentrations of Ca2�, Mg2�, Na�, K�, NH�4 , Clÿ, NOÿ3 , and SO2ÿ

4 were determined usingion chromatography (Dionex Model 2020).

RESULTS

Temperature, precipitation, snowpack and stream¯ow

Temperatures at 0 cm depth in the forest ¯oor (surface of mineral soil) remained above freezingthroughout winter (1 November to 30 April, the snowfall season) and averaged 0.4 8C. Temperatures in thesnowpack 10 cm above the forest ¯oor were almost constant to early April and ranged fromÿ0.3 toÿ1.4 8C.Precipitation amount (78 cm) for water year 1997 (1 October, 1996 to 30 September, 1997) was 2 cm belowthe long-term average, and winter precipitation amount (27.5 cm) was average (Figure 2) (Stottlemyer andToczydlowski, 1996b). For the 1997 water year, evapotranspiration was 53 cm or 68% of precipitation input.Most evapotranspiration (33 cm) occurred from mid-June to the end of October.

A continuous snowpack formed early December, and had completely melted by the end of April. Thepeak SWE was 24.5 cm the last week of March. Cumulative winter precipitation amount exceeded peak SWEby 3 cm.

Seasonal change for selected cations

Precipitation and snowpack Ca2� concentrations increased during the peak snowmelt period which beganthe week of 13 March and ended seven weeks later (Figure 3). Snowmelt Ca2� concentrations showedpossible evidence of a pulse the third week of melt, but generally declined throughout the snowmelt period.The Ca2� concentrations in shallow and deep lysimeter soil water were greater when soils were unsaturated,declined about 20% when soil water became saturated, and again increased to pre-snowmelt concentrations

Figure 2. Cumulative precipitation amount and streamwater discharge, and mean snowpack water equivalent (SWE) for wateryear 1997, Calumet watershed, Michigan



Figure 3. Weekly mean Ca2� concentration in precipitation, snowpack, snowmelt (station C5), soil water (station C5), andstreamwater, November 1996 to June 1997, Calumet watershed, Michigan



when surface soils became unsaturated. The pattern and magnitude of concentration changes were similar tosoil water in the well as water height approached the soil surface. During the ®rst half of snowmelt,streamwater Ca2� concentration declined from 980 to 425 meq Lÿ1.

Unlike Ca2�, the H� concentration in precipitation declined from February to May (Figure 4). Thesnowpack H� concentration also began to decline by middle February. During the initial days of the peaksnowmelt period, the snowmelt H� concentration peaked (pH � 4.3), then declined after the snowpack hadlost about 50% SWE. By 3 April, the snowmelt pH averaged about 4.98. The average H� in snowmelt was23 meq Lÿ1. Tension lysimeter soil water H� concentration (mean52 meq Lÿ1, pH 5.75) changed little priorto, during, and following snowmelt. Similarly, the H� concentration (mean 1 meq Lÿ1) in saturated soil waterwithin the well remained unchanged throughout the melt period with an average pH of 5.98. Winterstreamwater pH ranged from 7.7 on 7 January, to 7.15 (H� � 0.07 meq Lÿ1) on 6 April. The over-wintermean H� was 0.03 meq Lÿ1.

Precipitation NH�4 concentration (Figure 5) increased during winter similar to the trend for Ca2�.However, the snowpack NH�4 concentration varied little throughout winter. NH�4 concentration increasedin early snowmelt similar to the pattern for H�. In tension lysimeter soil water, NH�4 was only detectableafter peak snowmelt when the shallow soils again became unsaturated, but the concentration was50.5 meq Lÿ1 approaching the level of detection. The NH�4 concentration in saturated soil water in the wellsharply increased during the rise in soil water level. NH�4 concentration in streamwater remained below thelevel of detection after early February. For the 1997 water year, greater than 97% of precipitation NH�4input was retained in the ecosystem primarily in the forest ¯oor and shallow soils.

Seasonal change for selected anions

Precipitation NOÿ3 concentrations increased from December to early May in a pattern similar to Ca2� andNH�4 (Figure 6). The late winter snowpack NOÿ3 concentration increased similar to that for Ca2�. Thechanges in snowmelt NOÿ3 concentration were similar to H�. Unlike Ca2�, H�, and NH�4 , tension lysimetersoil water NOÿ3 concentrations increased when soil water became saturated. Saturated soil water in the wellshowed detectable NOÿ3 concentrations before peak snowmelt. NOÿ3 concentrations more than doubled asthe soil water level in the well increased, but concentrations rapidly decreased as soil water level declined.Streamwater NOÿ3 concentrations increased throughout winter, increased about 40% during initialsnowmelt, then gradually declined to near detection levels as stream¯ow approached base ¯ow.

Precipitation SO2ÿ4 concentrations (Figure 7) increased throughout winter in a pattern similar to Ca2�,

NH�4 , and NOÿ3 . But snowpack SO2ÿ4 concentrations showed little change after early winter, a pattern

similar to NH�4 . Snowmelt SO2ÿ4 concentrations increased during the peak melt period, but were highly

variable similar to H� and NOÿ3 . Unsaturated soil water SO2ÿ4 concentrations in shallow lysimeters declined

about 30% when soil water became saturated while SO2ÿ4 concentrations in deeper soil lysimeters gradually

increased. The SO2ÿ4 concentrations in saturated soil water in the wells and in streamwater were similar, and

showed a steady decline during and following peak snowmelt.Precipitation Clÿ concentration showed little change during winter, but increased near the end of the

snowpack season (Figure 8). Snowpack Clÿ concentrations varied the least of any ion (range 1.7 to3.2 meq Lÿ1). Snowmelt Clÿ concentrations doubled from 2 to 4 meq Lÿ1 during the peak melt period. Duringsnowmelt soil water Clÿ concentrations in shallow lysimeters showed a pattern identical to SO2ÿ

4 . The Clÿ

concentration in deep lysimeter soil water increased when soil water was unsaturated, then decreased460%when soil water became saturated. Clÿ concentrations in well water showed no change prior to, during, orfollowing snowmelt. Streamwater Clÿ concentrations declined during snowmelt in a pattern similar to Ca2�.

Streamwater HCOÿ3 , DOC, and Si

Streamwater HCOÿ3 concentrations were high throughout winter (Figure 9). During snowmelt, HCOÿ3concentrations decreased from 920 to 465 meq Lÿ1, a decline similar to Ca2� (980 to 425 meq Lÿ1). Dilutionof base cation (CB) concentration and the reduction of HCOÿ3 concentrations made up490% of ANC loss.



Figure 4. Weekly mean H� concentration in precipitation, snowpack, snowmelt, soil water, and streamwater, November 1996 toJune 1997, Calumet watershed, Michigan



Figure 5. Weekly mean NH�4 concentration in precipitation, snowpack, snowmelt, soil water, and streamwater, November 1996 to June1997, Calumet watershed, Michigan



Figure 6. Weekly mean NOÿ3 concentration in precipitation, snowpack, snowmelt, soil water, and streamwater, November 1996 to June1997, Calumet watershed, Michigan



Figure 7. Weekly mean SO2ÿ4 concentration in precipitation, snowpack, snowmelt, soil water, and streamwater, November 1996 to June

1997, Calumet watershed, Michigan



Figure 8. Weekly mean Clÿ concentration in precipitation, snowpack, snowmelt, soil water, and streamwater, November 1996 toJune 1997, Calumet watershed, Michigan



Streamwater DOC concentrations showed a unique trend during snowmelt. Concentrations increasedduring snowmelt, and peaked just after peak stream¯ow concurrent with soil water in the wells reaching theforest ¯oor. Concentrations then gradually declined with the decrease in stream¯ow. At peak stream¯ow,which was 30 times the over-winter average, DOC ¯ux was 65 times the over-winter average and the DOCconcentrations was 2.5 times the mean at base ¯ow.

Si concentrations were higher during winter and after snowmelt runo� (Figure 9). Concentrations declinedfrom 10 to 7.5 mg Lÿ1 during snowmelt in a pattern similar to Ca2� and HCOÿ3 , but the magnitude ofdecline was less (25%).

DISCUSSION

Hydrology

The peak SWE was several cm above the long-term average, but below the average of the early 1980s whenstudy began at Calumet. The average snowpack season is from late November to early April. In 1997, the

Figure 9. Weekly mean streamwater HCOÿ3 , DOC, and Si concentration, and streamwater discharge, November 1996 to June 1997,Calumet watershed, Michigan



snowpack existed two weeks longer. Since 1980, the average winter cumulative precipitation amount exceedspeak SWE by 7 cm, but in 1997 it was only 3 cm. The small SWE loss up to peak SWE re¯ects the constantfreezing temperatures deep in the snowpack, and few mid-winter thaws. Snowpack sublimation appears515% of precipitation input (Stottlemyer and Toczydlowski, 1991).

The rapid change in soil water level in the well (Figure 3) and the absence of signi®cant overland ¯ow(Stottlemyer and Toczydlowski, 1991) indicate how completely snowmelt entered the forest soil. Winterstream¯ow is correlated (p5 0.001, r2 � 0.5ÿ 0.7) to change in water level at both wells (Stottlemyer andToczydlowski, 1996b). Well water levels also show a diurnal pattern similar to stream¯ow, but with a peak®ve hours before stream¯ow. This time lag decreased as the watershed approached peak snowmelt. Theserelations show the importance of snowmelt in®ltration on seasonal soil water level and stream¯ow, and thesigni®cance of soil water contributions to annual stream¯ow (Kennedy et al., 1986). While precipitationthroughout the year was uniformly distributed (Figure 2), the decline of soil water level to 42 m depth bylate June indicates how important evapotranspiration by tree foliation is to the annual watershed hydrologicbudget. From mid-June to leaf fall, evapotranspiration was a major process limiting soil water recharge. Thelow soil water levels throughout the growing season explain why the stream rarely responds to summer rains(Stottlemyer and Toczydlowski, 1996b). After leaf fall, the reduced evapotranspiration and gain inthroughfall increased stream¯ow, but evaporation and gradual recharge kept cumulative stream¯ow belowprecipitation amount. Soil water in the wells did not reach the 52 m level until mid or late January.

Seasonal chemical change

The increase in late winter and early spring precipitation Ca2� and NH�4 concentration we attribute to lossof the regional snowpack and increased dust and aerosol formation and exposure of fertilized ®elds (Glassand Loucks, 1986). The seasonal increase in precipitation NOÿ3 and SO2ÿ

4 concentration was coincident witha shift in wind direction from the south, and is observed in other parts of the Great Lakes region (Semkinand Je�ries, 1988). In other portions of the northeastern US, as New England, and in eastern Canada,precipitation NOÿ3 and SO2ÿ

4 concentrations generally peak in summer (NADP 1982±1996).As soil water and streamwater levels declined, streamwater Ca2�, HCOÿ3 , and Si concentrations rapidly

increased, and Clÿ showed a slight increase (Figures 3, 8 and 9). Conversely, streamwater NH�4 , SO2ÿ4 , NOÿ3 ,

and DOC concentrations continued to decline or remained near detection levels (Figures 5, 6, 7 and 9). Theincrease in streamwater Ca2�, HCOÿ3 , and Si concentrations re¯ect greater relative contributions of soilwater from deeper ¯owpaths, increased contributions by mineral weathering at greater soil depths, andreduced soil water movement in the biologically active forest ¯oor and surface mineral soils. Streamwaterbase cation (CB), HCOÿ3 , and Si concentrations are greatest in late summer when stream¯ow is lowest.

Calcium was the dominant cation in surface water of the Calumet watershed, and change in this iondominates seasonal trends in CB concentrations (Stottlemyer and Toczydlowski, 1996b). The Ca2� con-centration and content in precipitation, the snowpack, and snowmelt (Figures 3 and 10) were insigni®cantcompared to streamwater even accounting for concentration increases with evapotranspiration. In unsatur-ated shallow soil water, Ca2� concentrations were high and steady re¯ecting a consistent and longerresidence time for soil and water during winter and the growing season, the high evapotranspiration rateduring the leaf-on period (Figure 2), and the large forest ¯oor and surface soil exchange capacity (D. Hanson,Dept. Biological Sciences, Michigan Technological Univ., unpub. data). The rapid recovery of Ca2�

concentration in shallow unsaturated soil water following peak snowmelt demonstrated how rapidly soilCa2� exchange can occur. When the shallow soil water became saturated, its change in Ca2� concentrationwas similar to that observed in well soil water. However, at the watershed level the decline in streamwaterCa2�, and HCOÿ3 , concentration was of greater magnitude and lasted much longer. The snow and soillysimeter plots and one well were located in the lower third of the watershed. The peak SWE signi®cantlyincreases with elevation in Calumet (Stottlemyer and Toczydlowski, 1996b), and the greater streamwater CB

dilution and HCOÿ3 reduction during snowmelt re¯ected this larger input. As observed in the growingseason, during the snowpack period streamwater Ca2�, HCOÿ3 , and Si concentrations were constant or



steadily increased, the product of increased contributions of soil water from deeper ¯owpaths and the morereadily weathered substrate with depth (Rice and Bricker, 1995). Over-winter increases in streamwater ionconcentration accentuated the e�ect of dilution on streamwater chemistry and ¯ux during snowmelt.

The change in H� concentration among ecosystem components was unique for the ions studied. Upto early March, snowpack H� concentrations were similar to precipitation. H� concentrations increasedin early snowmelt, but soil water H� concentrations were 510% snowmelt concentrations. The low H�

concentrations in both saturated and unsaturated soil water and the minor change in streamwaterconcentrations during snowmelt re¯ect the near complete H� retention by exchange in the forest ¯oor andshallow soil.

Similar to H�, precipitation and snowmelt NH�4 inputs were also strongly retained in the forest ¯oor andshallow soils. The peak snowpack NH�4 content was about 20 eq haÿ1 (Figure 10). However, inorganic Npools in the top 10 cm of the forest ¯oor and surface soil are almost an order-of-magnitude greater(Stottlemyer and Toczydlowski, 1996a). Forest ¯oor and surface soil net N mineralization rates are negativefromDecember through February and inMay. DuringMarch and April, net mineralization rates are positiveand average 17 eqN haÿ1 moÿ1. The steady forest ¯oor temperatures andmoisture content throughout winterpermit signi®cant amounts of N mineralization to occur. Past research shows microbial immobilization is themajor process regulating soil available N during winter. The lack of signi®cant soil water movement throughthe forest ¯oor and shallow soils during most of the year may also be a factor. Shallow unsaturated soil waternever showed NH�4 concentrations above detection limits, but soil water in the well showed a brief increase inNH�4 concentration concurrent with peak snowmelt rates, snowmelt NH�4 concentration, and rapid rise insoil water levels (Figure 5). The increase in soil water NH�4 could be from snowmelt, but the watershedinorganic N budgets suggest it was largely by exchange from soil available N pools.

As with NH�4 , the snowpack NOÿ3 content was small compared to soil inorganic N pool size (Stottlemyerand Toczydlowski, 1996a). In unsaturated lysimeter soil water, NOÿ3 concentrations were at the level ofdetection before and following peak snowmelt. Unsaturated soil water can have a long residence time. TheNOÿ3 ion is poorly adsorbed by soil exchange sites, but in the absence of signi®cant soil water movement theproducts of over-winter nitri®cation are rapidly immobilized by microbial activity. Evidence for this comesfrom the negative net NOÿ3 mineralization rates throughout winter except in March (Stottlemyer andToczydlowski, 1996b). The higher NOÿ3 concentrations in saturated soil water in the well prior to peaksnowmelt were likely the result of small but steady snowmelt moving over-winter forest ¯oor and shallow soilN mineralization products from the biologically active surface soils and forest ¯oor. Streamwater NOÿ3concentrations gradually increased during winter (Figure 6), but NOÿ3 ¯ux was small (Figure 10). Theincrease in soil net N mineralization and nitri®cation during winter at Calumet likely re¯ects reducedmicrobial immobilization because of a reduction in the amount of labile carbon. Reduced NOÿ3 immobili-zation likely explains the gradual increase in streamwater NOÿ3 concentrations during winter. Followingpeak snowmelt, the combined increase in microbial immobilization rates with warming soils (Stottlemyerand Toczydlowski, 1996a) and increased plant uptake quickly reduced NOÿ3 concentrations to detectionlimits in saturated soil water in the well and in streamwater.

Watershed SO2ÿ4 input and output from the beginning of the water year were balanced up to peak SWE

(Figure 10). SO2ÿ4 output began to exceed input during peak snowmelt. By middle June when stream¯ow

returned to base¯ow, SO2ÿ4 output was twice input. Most of the additional SO2ÿ

4 appeared to come from theforest ¯oor and surface soil. SO2ÿ

4 is the dominant anion in soil water at both the 15 and 30 cm depths(Stottlemyer and Toczydlowski, 1996b). Soil water SO2ÿ

4 concentrations at the shallow lysimeter depth werereduced a third when soil water became saturated, but they increased at the 30 cm depth (Figure 7). Soilwater SO2ÿ

4 concentrations were at least four times those in snowmelt. Saturated soil water in the well andstreamwater SO2ÿ

4 concentrations showed only a gradual decline during peak snowmelt. Sulfur mineraliza-tion rates would not be rapid enough to provide su�cient SO2ÿ

4 to o�set a large decline in streamwaterconcentration by dilution during snowmelt. Also, the Calumet surface bedrock is Freda sandstone, andcontains little or no sul®des (T. Bornhorst, Dept. Geology, Michigan Technological Univ., pers. comm.). The



Figure 10. Comparison of mean weekly snowpack ion content with cumulative weekly precipitation input and streamwater output frombeginning of 1997 water year, 1 October 1996, Calumet watershed, Michigan



high soil water and streamwater SO2ÿ4 concentrations and ¯ux during and following snowmelt must come

from soil desorption.Chloride concentrations should be the least altered by watershed mineralization or bilogical processes, and

seasonal trends in concentration among components may further explain ecosystem processes regulatingessential nutrients in snow dominated ecosystems. Similar to SO2ÿ

4 , watershed Clÿ inputs and outputs fromthe beginning of the 1997 water year were equal up to the time of peak SWE (Figure 10). By mid-June whenstream¯ow returned to base¯ow, Clÿ output exceeded inputs by 40%, half of which could be accounted forby snowpack Clÿ release with the remainder likely from dry deposition to the forest canopy the previousgrowing season (Cappellato and Peters, 1995) and, based upon our Clÿ concentrations in soil water duringsnowmelt, mineralization. Annual mean streamwater Clÿ concentration (19 meq Lÿ1) exceeds precipitation(4 meq Lÿ1). Evapotranspiration would account for about 10 meq Lÿ1 of the di�erence with the remaindercoming from mineralization. Streamwater Clÿ concentrations were greater during base ¯ow indicating mostmineralization occurs in soils deeper than 2 m or bedrock.

The seasonal trend in streamwater DOC concentration during snowmelt re¯ects the extent over-winterforest ¯oor and shallow mineral soil mineralization processes occur. Earlier more intensive lysimeter study(Stottlemyer and Toczydlowski, 1991) showed snowmelt contributed about 20% of total DOC ¯ux beneaththe forest ¯oor, the forest ¯oor about 75%, and surface soils about 5%. Except for snowmelt, these estimatesexclude the time when shallow soil water is saturated. Snowmelt DOC comes from dissolution of organicdebris in the snowpack. But the signi®cant DOC reservoir contributing to streamwater output is the forest¯oor. Over-winter mineralization processes in the unfrozen forest ¯oor and soil maintain this reservoir, andthe DOC is mobilized by percolating snowmelt and particularly by lateral ¯ow of saturated soil water atshallow depths. This seasonal increase in DOC ¯ux is important to the aquatic ecosystem. DOC containsnumerous labile carbon forms, a seasonally important energy source to the aquatic ecosystem.

In sum, winter precipitation stored in the snowpack had several e�ects on seasonal watershed chemistry.The snowpack insulated the forest ¯oor and soil against freezing which permitted over-winter mineralizationand other biological processes to maintain shallow subsurface ion and DOC reservoirs. Small, but steady,snowmelt throughout winter removed some readily mobilized products, as NOÿ3 , which resulted in highover-winter streamwater concentrations but very small ¯ux. Winter soil water levels and ¯owpaths were deepwhich increased soil water and streamwater CB , HCOÿ3 , and Si concentrations. In spring, rapid snowmeltincreased soil water levels which increased removal of ions and DOC from the biologically active forest ¯oorand shallow soils. The snowpack solute content was a minor factor in regulating streamwater ion concen-tration or ¯ux during and following peak snowmelt. Exchangeable ions, weakly adsorbed anions, and DOCin the forest ¯oor and surface soils dominated the chemical concentrations and ¯ux in soil water, andstreamwater. Following peak snowmelt, soil microbial immobilization and rapidly increased plant uptake oflimiting nutrients removed nearly all available nitrogen from soil water and streamwater. During the growingseason up to leaf-fall, high evapotranspiration increased subsurface ¯owpath depth which in turn removedweathering products from deeper soils and increased streamwater CB , HCOÿ3 , and Si concentrations.

ACKNOWLEDGEMENTS

This research was supported by the Watershed Research Program, Biological Resources Division, US Geo-logical Survey. The authors acknowledge the Lake Superior Land Division, Champion International, whichhas kindly permitted access and use of the Calumet watershed.

REFERENCES

Caine N. 1989. Hydrograph separation in a small alpine basin based on inorganic solute concentrations. Journal of Hydrology 112:89±101.

Cappellato R, Peters NE. 1995. Dry deposition and canopy leaching rates in deciduous and coniferous forests of the Georgia Piedmont:an assessment of a regression model. Journal of Hydrology 169: 131±150.



Eichenlaub L, Harmon JR, Nuenberger FV. 1990. Climatic atlas of Michigan. Univ. Notre Dame Press: Notre Dame, Indiana: 165pp.Glass G, Loucks O. 1986. Implications of a gradient in acid and ion deposition across the northern Great Lakes states. Environmental

Science and Technology 20: 35±43.Harrington R, Bales RC. 1998. Interannual, seasonal, and spatial patterns of meltwater and solute ¯uxes in a seasonal snowpack.

Water Resources Research 34(4): 823±831.Hazlett PW, English MC, Foster NW. 1992. Ion enrichment of snowmelt water by processes within a podzolic soil. Journal of

Environmental Quality 21: 102±109.Je�ries DS, Cox CM, Dillon PJ. 1979. Depression of pH in lakes and streams in Central Ontario during snowmelt. Journal of the

Fisheries Research Board of Canada 26: 640±646.Johannessen M, Henriksen A1978. Chemistry of snow meltwater: changes in concentration during melting.. Water Resources Research

14: 615±619.Kattelmann RC. 1989. Spatial variability of snow-pack out¯ow at a site in the Sierra Nevada. Annals of Glaciology 13: 124±128.Kelso JRM, Minns CK, Lipsit JH, Je�ries DS. 1986. Headwater lake chemistry during the spring freshet in North Central Ontario.

Water Air Soil Pollution 29: 245±259.Kennedy VC, Kendall C, Zellweger GW, Wyerman TA, Avanzino RJ. 1986. Determination of the components of storm ¯ow using

water chemistry and environmental isotopes, Mattole River Basin, California. Journal of Hydrology 84: 107±140.NADP. 1982±1996. NADP NTN annual data summaries: precipitation chemistry in the United States, 1982±1991. National Atmospheric

Deposition Program, Natural Resources Ecology Laboratory, Colorado State Univ.; Ft. Collins, CO.Pierson DC, Taylor CH. 1985. In¯uence of snowcover development and ground freezing on cation loss from a wetland watershed

during spring runo�. Canadian Journal of Fisheries and Aquatic Science 42: 1979±1985.Rice KC, Bricker OP. 1995. Seasonal cycles of dissolved constituents in streamwater in two forested catchments in the mid-Atlantic

region of the eastern USA. Journal of Hydrology 170: 137±158.Semkin RG, Je�ries DS. 1988. Chemistry of atmospheric deposition, the snowpack, and snowmelt in the Turkey Lakes Watershed.

Canadian Journal of Fisheries and Aquatic Science 45(Supp. 1): 38±46.Stottlemyer R. 1987. Snowpack ion accumulation and loss in a basin draining to Lake Superior. Canadian Journal of Fisheries and

Aquatic Science 44(11): 1812±1819.Stottlemyer R, Toczydlowski D. 1991. Stream chemistry and hydrologic pathways during snowmelt in a small watershed adjacent

Lake Superior. Biogeochemistry 13: 177±197.Stottlemyer R, Toczydlowski D. 1996a. Modi®cation of snowmelt chemistry by forest ¯oor and mineral soil, Northern Michigan.

Journal of Environmental Quality 25: 828±836.Stottlemyer R, Toczydlowski D. 1996b. Precipitation, snowpack, stream-water ion chemistry, and ¯ux in a northern Michigan

watershed, 1982±1991. Canadian Journal of Fisheries and Aquatic Science 53: 2659±2672.Wiener JG, Eilers JM. 1987. Chemical and biological status of lakes and streams in the Upper Midwest: Assessment of acidic

deposition e�ects. Lake Reservoir Management 3: 365±378.



seasonal change in precipitation, snowpack, snowmelt, soil water and streamwater chemistry, northern...

Documents