VARIATIONS IN THE CONCENTRATION OF CHEMICAL CONSTITUENTS OF A
STREAM WATER DURING THE
SNOWM肌T SEASONKeisuke SUZUKI
Abstract Variations of the water quality of a small stream during the snowmelt season were studied． The discharge and concentration of the chemical constituent in the stream were also simulated． The concentrations of chemical constituents， except Cl， in the stream water were diluted by meltwater during the snowmelt season． The Cl concentration was higher during the snowmelt season than before and after the snowmelt season and was higher at the
early snowmelt and decreased as the snowmelt progressed． In the studied watershed only small amounts of Cl are released from the rocks， and the predominant source of Cl in the stream water is precipitation． The variation of Cl concentration in the stream water was assumed to be infiuenced by Cl concentration in meltwater． This variation was simulated by using a runoff model and a model of meltwater chemistry．
1．Introduction
The early meltwater f［owing from the snow cover contains a high concentration of chemical constituents， and the concentrations of those constituents in the meltwater decrease as the snowmelt progresses（Johannessen and Henriksen，1978；Suzuki，1982）．
Stream water contains chemical constituents which are released not only from the snow cover but also from the rocks composing the watershed． The qualities of the meltwater are di ffe rent before and after contact with the rock． The concentration of the chemical con−
stituents in stream water is influenced by the snow cover and the lithology of the rocks in the snowmelt season． However， chemical constituents which are not significantly released from rocks are influenced by only the chemica1 characteristics of meltwater．
Variations during the snowmelt season in the water quality of a small stream were studied． The discharge and concentration of the chemical constituent in the stream were also simulated，
2．Study Watershed
The study watershed is located to the west of Sapporo（Fig．1）． The watershed has an area of O．683㎞2．The bedrock of the upper part of the watershed consists of hornblende
137一
ボSaPP°「°
＄
0 2km
目
歩
0 200 m
Fig。1 Location of study watershed
bearing augite hypersthene andesite， and the lower part consists of propylite．The watershed is entirely covered with forest vegetation dominated by、Acer mono var． glabrum and Tilia laponlca・
3．Observational Methods
The water level was continuously measured by means of a horizontal float recorder． The discharge was calculated by using the water level−discharge curve． Stream water samples were taken with polyethylene bottles． Each sample was filtered through a LO micron pore size membrane filter which was previously washed with distilled water． The chemistry of stream water was analyzed in much the same way as Suzuki（1982）． However， SiO2 was added to list of elements searched for． The SiO2 content was determined by spectrophotometry．
The depth of the snow cover was measured near the outlet of the watershed． Samples of the total snow cover were taken in order to estimate the amount of chemical constituents in the snow cover． Sampling and analytical procedures for the snow cover were the same as Suzuki（1982）．
4．Water Quality of the Stream Water during the Snowmelt Season
The runoff is shown in Fig．2．The runoff was obtained by dividing the daily discharge by the basin area． The runoff was constant until March 28，1980． This runoff was equivalent to the baseflow fbr the winter． The snowmelt runoff began on March 29，1980， and two peaks were observed on April 5 and 20，1980．
40
宕30
ε 竃 20 至
10
0 20
1980Mar．
30
Apr．
10 20 30
May
10 20 30
Fig．2 Runoff for the period from March 19 to May 31，1980
The water qualities of the stream are shown in Fig．3．Since the conductivity decreased during the snowmelt season， the meltwater diluted the concentration of the dissolved load in the stream water． The pH decreased also，but the concentration ofH＋calculated from the pH increased during the snowmelt season． The concentrations of all chemical constituents except Cl decreased during the snowmelt season． The CI concentration was higher during the snowmelt season than befbre and after the snowmelt season． The Cl concentration was higher at the early snowmelt and decreased as the snowmelt progressed．Only small amounts of CI are released from the rocks， and the predominant source of CI in the stream water is precipitation．The other elements are supplied from the rock as the water flows down．
The concentrations of the chemical constituents in the snow cover at the study area are shown in Table 1． Based on this result， we will estimate the amounts of elements added from precipitation to the stream water． The source of Cl in the stream is assumed to be precipitation， and the ratios of other elements to Cl in the precipitation can be obtained from Table l．The ratios of precipitation input in the base flow are estimated as shown in Table 2． A large portion of each element is released from the rock． During the snowmelt season， the dilute effect influences the concentrations of the chemical constituents except the C1， and the concentrations decrease，The correlation matrix of the various water qualities is given in Table 3． The concentrations of the chemical constituents except Cl have a negative correlation with the discharge（Q）；the Cl concentration does not relate to the discharge．
139
（ζ0ε）と
＝O
2 t t
ム「5
サ
7777
432」0 0000
Si
λ25
13 12 11 10
9 8
20
15 10
3．0 2．5 2．0
10
9 8 7 130 110
90 70
（言εこ
ミOE）可Z
10 20 30 10 20 30 10 20 30
1980Mar． Apr． May
Fig．3 Water qualities in stream water for the period from March to May，1980
Table 1 Concentration of chemical constituents ln SnOW COVer
Table 2． Ratio of precipitation input in the base flow
Concentration
（mg／1）
（％）
Na KMg Ca Cl SO4
2．07 0．34 0．26 0．21 3．7 2．1
MK晩頓so 542 249178
Tab le 3 Correlation matrix of water qualities
Q（m3！、ec）
Cond．
（μS／cm）
Na
（mg／1）
K
（mg／l）
Mg
（mg11）
Ca
（mg！1）
Cl
（mg／1）
SO4
（mg／l）
ミOE）
18 16 14
り 12 10
0．4
0．3
O．2
0．1
O
1979 10Mar．
20 30
Apr．
10 20 30
Fig．4 Discharge and CI concentration for the period from March 5 to April 30，1979
The same variation of CI concentration was also observed in the snowmelt season of 1979．The discharge and Cl concentration in 1979 are shown in Fig．4． The CI concentration was higher at the early snowmelt and decreased as the snowmelt progressed． The such a variation of the Cl concentration during the snowmelt season is concluded to be a general
＿141一
phenomenon． Since the source of Cl is the snow cover， the variation of the Cl concentration in the stream water is affected by the processes of snowmelt．
5．Simulation of Cl Concentration in the Stream Water during the Snowmelt Season
The variation of the Cl concentration in the stream water during the snowmelt season is considered to be defined by the variation of the Cl concentration in the meltwater． We will s㎞ulate the variation of the Cl concentration in the stream water using the model of meltwater chemistry described in Suzuki（1982）．
Model for estimating the runoff
First a tank・model（Sugawara，1972）is used to simulate the runoff． The meltwater is the input， and the runoff is the output． Since the snowmelt varies diurnally， the runoff is calculated daily．
It is difficult to successively observe the amount of meltwater in a mountainous water・
shed， and therefbre th6 da丑y amount of meltwater must be estirnated in some way． We est㎞ated the daily amount of meltwater f士om the daily accumulated temperature． The daily accumulated temperature is obtained from the hourly temperatures above O°C． The relation between the depth of the snow cover and its water equivalent during the snowmelt season at the Agricultural Experiment Farms of Hokkaido University is shown in Fig．5．
The density of the snow cover obtained by dividing the water equivalent by the depth
0 0 4
（∈ε）
0
●
・ ・
●
●
●
●●
●●
●
○
●
●
0 20 40 60
Depth of snow cover （cm）
80
Fig。5 Relation between the depth of snow cover and the water equivalent of snow cover during snowmelt season at the Agricultural Experi・
ment Farms of Hokkaido University
30
00 100 200
Accumulated temperature （°C，day）
Fig．6 Relation between the daily amount of snow−
melt and the daily accumulated temperature at the study site
of the snow cover is seen to be constant during the snowmelt season。The relation between the daily amount of snowmelt at the study site obtained by multiplying the mean density of the snow cover by the daily reduction of its depth and the daily accumulated temperature is shown in Fig．6． Hourly temperatures at the study site were obtained by subtracting O．8°Cfrom the hourly temperature at the Sapporo District Meteorologica10bservatory．
The regression equation of the plots in Fig．6is
SM＝0．08・AT＋10．2 （r＝0．90）＿＿＿．＿．．．＿．．＿．（1）
SM daily amount of snowmelt AT daily accumulated temperature．
The second term on the right side accounts for factors contributing to the snowmelt other than temperature． If the water equivalent of the snow cover before snowmelt is estimated from equation（1）， the estimated water equivalent exceeds the actual water equivalent． In order to make the estimated value coincide with the actual value， equation（1）
is rewritten as follows：
SMニ0．08・AT＋7．4 ．．．．．．．．．．．．．．．．．．．．．．．．．．．．．．．．．．（2）
The daily amount of snowmelt can be estimated by using equation（2）． Because air tem−
perature decreases with altitude， the daily amount of snowmelt was estimated at 50 m intervals． The lapse rate is assumed to be o．3°c／50m． The amount of precipitation was added to the snowmelt when the air temperature at the Sapporo District Meteorological Observatory rose above 3°C． The result of the estimation is given in Fig．7．
The tank−model is determined with the estimated snowmelt input and the measured
143
50
40 バE
）30E
ご ゆ
ε20≧
oに
の 10
0
1980 20
Mar．
30 10 20 30 10 Apr． May
Fig．7 Estimated daily amount of snowmelt
20 30
S．M．
sat
x1↑
h轟
」
Y2→Yl→
O
a12 17ΨX
↑
b1
◎
Z2vX3
bo
C1
Y3→
45
0．1
匝
Y→
0．15
0．005
Fig．8 Tank−model used for the simulation
nmoff output． The tank・model was set up as on the left side of Fig．8． Each tank corre−
sponds to surface runoff， interflow， and base flow， respective豆y．
50
＾40E
∈
） 30 と
o⊆
匡コ20
10
e．．e．ee ■te
0 20
1980Mar．
● ●3
《〜、礁ノ
30 10
Apr．
、・ ．
20
鵜暴ヤ・・
30 10 May
measured estimated
20 30
Fig．9 Estimated and measured runoff
The runoff is a function of the factors of each outlet． The factors were determined by the method of trial and error． The determined factors are given on the right side of Fig．8．
The est㎞ated runoff and measured runoff are shown in Fig．9． The correlation coefficient is O．96． The measured runoff can therefore be reproduced by this model， since it was confirmed by comparison of the esttmated runoff with the measured runoff．
Simulation of the Cl concentration in the stream water
The variation of the Cl load of the snow cover during the snowmelt season can be approximated with a quadratic equation（Suzuki，1982）． Continuous measurement of the Cl load of snow cover in a mountainous area is difficult；therefbre the Cl load ofthe snow cover was obtained by using the model． In order to approximate the variation of the Cl load of the snow cover with a quadratic equation， three conditions are required． These are，1）
the Cl load of snow cover at the beginning of the snowmelt，2）the Cl load is zero at the end of snowmelt， and 3）the reduction rate of the Cl load is zero at the end of the snowmelt．
The estimated Cl load under these conditions and the measured Cl load at the Agricultural Experiment Farms of Hokkaido University are given in Fig．10． The measured Cl load can be reproduced by this estimation． The reduction of CI load of the snow cover is assumed to be equivalent to the CI outflow from the snow cover． The estimated Cl outflow is shown in Fig．11．
The amount of Cl input to the runoff model is assumed to be equal to the amount of Cl outflow from the snow cover． The amount of Cl is added to the runoff model， and the Cl concentrations in each tank are determined by dividing the amount of Cl by the amount of water stored in each tank． The buffering effect is also considered． Ohba（1977）noted that the buffering mechanism was mainly due to the adsorption・released mechanism． Ten percent of the stored Cl is assumed to be adsorbed in each tank． The Cl concentration・in the base
145
（N∈り、oユ》
一〇
200
150
100
50
O
、、 、、
、
、、
ら b
、
㌦ も、
㌦、
ら覧
●、
、
駈●
、 、 覧 覧 魅 、
、 ●
＼●
、、
、 、 s 、 、 s s s 、 、 、
、 9、
、、
●㍉、A ●、、、
●、、●、
℃・．◎
㍉●㍉Q唱亀．●
開●
1980 20
Mar．
25 30
Apr．
5 10
Fig．10 Cl load of the total snow cover during the snowmelt season at the Agricultural Experiment Farms of Hokkaido University
100
80
ぐE
tt 60
0ユ 40）
一り 20
0 20
1980Mar．
30
Apr．
Fig・11
OOOoo
10 ． 20 30 10 20
May
Estimated Cl outflow from the snow cover
30
flow is assumed to be 8．5 mg／1． Based on these assumptions， the Cl concentration in the stream water can be estimated． The estimated Cl concentrations and measured Cl concen−
trations are given in Fig．12． The measured Cl concentration in stream water can be reproduced by this method of estimation．
14
12
@ 10
（ζ0ε
一〇
8
20
1980Mar．
八．．．．
30
Apr．
10 20 30 10
May
measured estimated
20
Fig．12 Estimated and measured Cl concentrations in the stream water
30
6．Conclusions
Alarge portion of the chemical constituents， except for Cl， in the stream water Was found to be released from rocks composing the watershed， and the concentrations of chemical constituents were diluted by meltwater during the snowmelt season． On the other hand，
because the source of Cl was found to be the meltwater， the Cl concentration in the stream water was higher at early snowmelt and decreased as the snowmelt progressed． The variation of Cl concentration in the stream water was assumed to be influenced by Cl concentration in meltwater． In order to verify this hypothesis， a runoff model was constructed， and the Cl concentration in the stream water was estimated by using this model． Since the measured Cl concentration in the stream water was reproduced by this estimation， the hypothesis was confirmed．
Acknowledgements
The author wishes to express his thanks to Professor Hiroshi Kadomura， Graduate School of Environmental Science， Hokkaido University， for his encouragement through this study．
Gratitude is also expressed to Dr． Masao Uziie， Faculty of Agriculture， Hokkaido University，
for the use of the atomic−absorption spectrophotometer．
This paper is dedicated to Professor Takamasa Nakano on the occasion of his retirement from Tokyo Metropolitan University．
147
References Cited
Johannessen， M． and A． Henriksen（1978）：Chemistry of snow meltwater：changes in concentration during melting。 Wat． Resour． Rθ∫．，14，615−619．
Ohba， T．（1977）：Simulation of Cl concentration of st正eam water using a watershed model． Geogr． Rθレ．
ノbρan，50，675−688．（in Japanese with English abstract）
Sugawara， M．（1972）：Ryushutsu Kaiseki、Ho rRunoff．4 nalyses7． Kyoritsu Shuppan， Tokyo，257p．
（in Japanese）
Suzuki， K．（1982）：Chemical changes of snow cover by melting．ノOp．」． Limnology，43，102−ll2．
