- Email: [email protected]
Overview of a simple model describing variation of dissolved organic carbon in an upland catchment
E(OLOGI(IIL mODELLInG Ecological Modelling 86 (1996) 183-188
ELSEVIER
Overview of a simple model describing variation of dissolved organic carbon in an upland catchment Elizabeth W. Boyer a
a,b,* George
M. Hornberger Diane McKnight b
a
Kenneth E. Bencala
b
Department of Environmental Sciences, University of Virginia, Charlottesville, VA 2290, USA b U.S. Geological Survey, MS496, 345 MiddlefieM Road, Menlo Park, CA 94025, USA
Abstract
Hydrological mechanisms controlling the variation of dissolved organic carbon (DOC) were investigated in the Deer Creek catchment located near Montezuma, CO. Patterns of DOC in streamflow suggested that increased flows through the upper soil horizon during snowmelt are responsible for flushing this DOC-enriched interstitial water to the streams. We examined possible hydrological mechanisms to explain the observed variability of DOC in Deer Creek by first simulating the hydrological response of the catchment using TOPMODEL and then routing the predicted flows through a simple model that accounted for temporal changes in DOC. Conceptually the DOC model can be taken to represent a terrestrial (soil) reservoir in which DOC builds up during low flow periods and is flushed out when infiltrating meltwaters cause the water table to rise into this "reservoir". Concentrations of DOC measured in the upper soil and in streamflow were compared to model simulations. The simulated DOC response provides a reasonable reproduction of the observed dynamics of DOC in the stream at Deer Creek. Keywords: Carbon; Hydrology; River ecosystems; TOPMODEL
1. I n t r o d u c t i o n A quantitative understanding of the factors controlling the variation of dissolved organic carbon (DOC) in headwater streams is needed for a number of reasons, not the least of which is because D O C interacts strongly with other dissolved substances (heavy metals in particular) and plays an important role in the transport of contaminants. In Deer Creek near Montezuma, Colorado, measurements of dissolved organic carbon (DOC) * Corresponding author.
from July 1991 to August 1992 show rapid decreases in concentration from a peak very early in the snowmelt period (on the ascending limb of the hydrograph); the peak concentration of D O C in the stream occurs well before the peak discharge. A m o n g the main factors controlling D O C variation in such headwater streams are the hydrological catchment responses, including flow paths through and residence times of water in the catchment. Models of the relationships between hydrological flows and D O C transport are important for quantifying the hypothesized mechanisms and thereby indicating whether observations are consistent with hypotheses.
0304-3800/96/$15.00 © 1996 Elsevier Science B.V. All rights reserved SSDI 0304-3800(95)00049-6
184
E. IV.. Boyer et al. / Ecological Modelling 86 (1996) 183-188
We examined possible hydrological mechanisms to explain the observed variability of D O C in D e e r Creek, by first simulating the hydrological response of the catchment using T O P M O D E L and then routing the predicted flows through a simple model that accounted for temporal changes in DOC. Conceptually the D O C model can be taken to represent a terrestrial (soil) reservoir in which D O C builds up during low flow periods and is flushed out when infiltrating meltwaters cause the water table to rise into this "reservoir". To evaluate the model, D O C concentrations were measured in the u p p e r soil and the stream at several locations. Water sampled from lysimeters in the u p p e r soil showed an accumulation of D O C during periods of low flow and a pronounced decline in D O C concentrations during snowmelt. These results, along with the general fidelity with which variations in stream D O C are captured using the model, support the hypothesized flushing mechanism.
to weekly later in the summer, for determination of D O C concentration. 3.2. Hydrological modeling
We applied the hydrological model T O P M O D E L (Beven and Kirkby, 1979) to simulate the catchment hydrology. The features of the U V A - U S G S version of the model that we use have been described by Wolock (1993). The topographic index which is at the core of TOPM O D E L , l o g e ( a / t a n f l ) , was calculated from a) Low flow conditions Litter l a y e r \ Upper soil horizon\ ~ Lower soil Soil water zone ~
~
The site of the study is the D e e r Creek catchment, located in Summit County, Colorado near Montezuma. The catchment is mountainous, ranging in elevation from about 3350 m to 4120 m, and drains an area of 10.6 km 2. Approximately half of the catchment is above the tree line.
3. M e t h o d s
~
..y,~'~~
Water
.... , T °\Wotor " °'tob,o
......
= s,2= .
2. S i t e d e s c r i p t i o n
,
Aquifer Groundwater
~~G1.~:~et:rmovernent
zone
b) High flow conditions Litter l a y e r ~ Upper soil horizon~ __.'~_,, Lowereoll horizon,~ ~ Water table r i d g e ~
~ ~
~"
~ -~
Water
3.1. Field instrumentation
Details of the field m e a s u r e m e n t procedures are given by Boyer (1993). Briefly, a stream gage was installed at the mouth of the catchment that provided a stage record from which the daily discharge was computed. Tension lysimeters, which provide water samples from the u p p e r soil, were installed at several locations on hillslope transects. W a t e r samples from the stream and soil lysimeters were collected, at a frequency ranging from daily near the beginning of snowmelt
" ' ' ' ' ' " ~ ~
A q u i f • r
'~"~---. Water rnovement
Groundwater zone
"Gravel in strearnbed
Fig. 1. Hydrological flushing of soil water. (a) During periods of low flow, subsurface flow is through lower soil horizons. (b) During high-flowconditions (spring snowmelt), the water table rises and flow to the stream is through the upper soil horizons as well. (Taken from Hornberger et al., 1994.)
E. W. Boyer et al. / Ecological Modelling 86 (1996) 183-188
D E M data as described by Wolock (1993). Average soil depth was assumed to be 1 m; an " u p p e r horizon", the presumed reservoir for high D O C levels, was assumed to be 0.5 m thick. TOPM O D E L was calibrated for Deer Creek using the Rosenbrock optimization algorithm. (See Hornberger et al., 1985, for a description of the use of the Rosenbrock method in conjunction with T O P M O D E L . ) The parameters optimized were two soil parameters, one intended to represent the rate of decrease of hydraulic conductivity with depth divided by porosity and the other the hydraulic conductivity of the surface soil (see Beven and Wood, 1983), and two snowmelt parameters, one the threshold temperature for melting and the other the coefficient relating the maximum rate of melting to the average daily temperature (see Bras, 1990).
185
snowmelt (Fig. 1). This would suggest that D O C in the vadose zone might build up during periods of low flow (through leaching from overlying organic-rich litter and through microbial activity) and that this D O C would be pushed out into the stream during periods of high flow. Such a conceptual model would indicate that concentrations of water flowing from the soil to the stream should be high in the initial part of a rainstorm or snowmelt event and then decrease as meltwaters continue to flush the zone formerly above the water table but now exposed to significant streamward flow of water. The simplest mathematical representation of flushing such as envisioned in the conceptual model is from a continuously stirred tank reactor (CSTR). In response to the influx of a flushing solution, such a simple device produces an exponential decline in concentration in the outflow solution with time. We constructed a simple mixing model, based on a CSTR, to explain temporal variation of D O C concentrations in the Snake River. We envision two subsurface reservoirs: one representing the " u p p e r soil" and one representing the "lower
3.3. Chemical modeling The conceptual model is that D O C in riparian and hillslope soils is "flushed out" by the formation of a ground-water ridge during periods of
Simulation o f DOC in Deer Creek
o
O0
~_3
o
e~ 2
•
~o
0 11-Jul-91
09-Sep-91
08-Nov-91 - -
07-Jan-92
simulated DOC
0"/-Mar-92 06-May-92 e
05-Jul-92
observed DOC
Fig. 2. Simulatcd and observed values of D O C for Deer Creek.
03-Sep-92
E. W. Boyer et al. / Ecological Modelling 86 (1996) 183-188
186
soil". Flows through the two subsurface reservoirs are apportioned on the basis of results from T O P M O D E L . We use a simple proration to apportion flow to the u p p e r reservoir. We assume that D O C in the lower reservoir is constant. From data we estimate that the D O C in the lower reservoir is 1.2 mg 1-1. To account for build-up of D O C in the soil, we use the model of Grieve (1991). D O C in the u p p e r reservoir varies according to: DOCupper(t
D O C s t r e a m = (FlOWoverland * D O C s n o w + FIOWuppe r * D O C u p p e r
+ Fl°wl . . . . * D O C I . . . . )/FlOWtotal
4. Results
+ 1)
= D O C u p p e r ( t ) + a * 10 0"04T
- (1 - e - k r ) * DOCupper(t ) -
The D O C in the stream is calculated as a simple mixture of waters from direct snowmelt and from the upper and lower soil reservoirs:
(1 - qupper( t ) / Vupper( t ) ) * D O C upper( t )
where t indicates time, T is temperature, a and k are rate constants for evolution and decay of D O C (see Grieve, 1991), qupper and Vupper are, respectively, the flow through and wetted volume of the u p p e r soil reservoir, and the time step is one day. Although the appropriate t e m p e r a t u r e is a soil temperature, we used air t e m p e r a t u r e as a surrogate. We used a = 0.11 and k = 0.002.
The simulated hydrological response for 19911992 was used as input to our simple D O C model. The simulated D O C response provides a reasonable reproduction of the observed dynamics of D O C in the stream at D e e r Creek (Fig. 2). This was expected from the results presented by Hornberger et al. (1994) who successfully applied this model to describe D O C dynamics in the Snake River, a stream that is confluent with D e e r Creek immediately downstream of our study reach. We attribute the apparent "delay" in the simulated D O C response to the fact that the initial time of rise is late in the simulated discharge hydrograph;
Simulation o f D O C in upper soil 35
30
25
2o r~ 15
10 f I l-Jul-91
09-Sep-91
08-Nov-91
07-Jan-92
07-Mar-92
06-May-92
05-Jui-92
Fig. 3. Simulated values of DOC in the upper soil reservoir.
03-Sep-92
E.W. Boyer et al. /Ecological Modelling 86 (1996) 183-188
187
5. Conclusions
presumably, the t e m p e r a t u r e index model calculates the initiation of significant snowmelt to be later than when the melt is actually initiated. The model captures the essence of the sharp decline in D O C concentrations in the stream after a peak early in the melt hydrograph. The real test of the modeling approach is whether the simulated D O C in the hypothetical soil reservoir was in any sense "close to" that measured in soil solution. Simulated D O C in the u p p e r soil reservoir ranged from about 5 mg 1-1 to about 35 mg 1-1 with p e a k concentrations preceding the initiation of the snowmelt runoff (Fig. 3). Peak measured concentrations of D O C in soil lysimeters ranged up to more than 80 mg 1-1 but most showed peak concentrations in the 20 to 50 mg 1-~ range (Boyer, 1993). Results from a riparian lysimeter, which may be most appropriate for conditions that we are simulating, show D O C concentrations that are very similar to those simulated (Fig. 4).
Neither the simulated discharges from T O P M O D E L nor the simulated D O C values from the C S T R model can be considered to be even nearly perfect representations of the data. Nevertheless, we conclude that the model results are broadly consistent with the available data collected in the stream and in the upper soil of the D e e r Creek catchment. Given the available data, we cannot reject the simple model as a valid description of processes occurring in the D e e r Creek system. The m e a s u r e m e n t s of soil D O C from the field sampling program in 1991 and 1992 lend credence to the model over and above that conferred from the simulation of the stream dynamics. As is often the case, examination of existing data in light of the modeling results suggests additional work that is necessary to more rigor-
D O C in an upper soil lysimeter 50
0.8
40
0.6
~30 o
20
10
0.4 .~
/
0.2
0 06-Apr-92
0 06-May-92
05-Jun-92 o
05-Jul-92 DOC
- -
04-Aug-92
03-Sep-92
03-Oct-92
discharge
Fig. 4. Observed concentrations of D O C in a shallow riparian soil. Stream discharge is shown to indicate the timing of the snowmelt event.
188
E.W.. Boyer et al. /Ecological Modelling 86 (1996) 183-188
ously test t h e h y p o t h e s e s e m b e d d e d in th e m o d e l a nd to r e f i n e t h e m o d e l itself so that c o m p a r i s o n s b e t w e e n m o d e l results and o b s e r v a t i o n s are rend e r e d less a m b i g u o u s . In particular, th e spatial a nd t e m p o r a l v a r i a t i o n s o f s n o w m e l t a n d soil p r o p e r t i e s , in a d d i t i o n to t o p o g r a p h y , should be c o n s i d e r e d in t h e r e f i n e m e n t o f th e c o n c e p t u a l model.
Acknowledgements This w o r k was s u p p o r t e d in p a r t by t h e N a tional S c i e n c e F o u n d a t i o n t h r o u g h G r a n t E A R 9304794.
References Beven, K.J. and Kirkby, M.J., 1979. A physically based, variable contributing area model of basin hydrology. Hydrol. Sci. Bull., 24: 43-69.
Beven, K.J. and Wood, E.F., 1983. Catchment geomorphology and the dynamics of runoff contributing areas. J. Hydrol., 65: 139-158. Boyer, E.W., 1993. Hydrology and the variation of dissolved organic carbon in soil and stream waters of two headwater catchments; Summit County, Colorado. M.S. Thesis, Department of Environmental Sciences, University of Virginia. Bras, R.L., 1990. Hydrology: an Introduction to Hydrologic Science. Addison-Wesley, Reading, MA, 643 pp. Grieve, I.C., 1991. A model of dissolved organic carbon concentrations in soil and stream waters. Hydrol. Processes, 5: 301-307. Hornberger, G.M., Beven, K.J., Cosby, B.J. and Sappington, D.E., 1985. Shenandoah watershed study: calibration of a topography-based, variable contributing area hydrological model to a small forested catchment. Water Resour. Res., 21: 1841-1850. Hornberger, G.M., Bencala, K.E. and McKnight, D.M., 1994. Hydrological controls on dissolved organic carbon during snowmelt in the Snake River near Montezuma, Colorado. Biogeochemistry, 25: 147-165. Wolock, D.M., 1993. Simulating the variable-source-area concept of streamflow generation with the watershed model TOPMODEL. U.S. Geological Survey Water-Resources Investigations Report 93-4124.
ELSEVIER
Overview of a simple model describing variation of dissolved organic carbon in an upland catchment Elizabeth W. Boyer a
a,b,* George
M. Hornberger Diane McKnight b
a
Kenneth E. Bencala
b
Department of Environmental Sciences, University of Virginia, Charlottesville, VA 2290, USA b U.S. Geological Survey, MS496, 345 MiddlefieM Road, Menlo Park, CA 94025, USA
Abstract
Hydrological mechanisms controlling the variation of dissolved organic carbon (DOC) were investigated in the Deer Creek catchment located near Montezuma, CO. Patterns of DOC in streamflow suggested that increased flows through the upper soil horizon during snowmelt are responsible for flushing this DOC-enriched interstitial water to the streams. We examined possible hydrological mechanisms to explain the observed variability of DOC in Deer Creek by first simulating the hydrological response of the catchment using TOPMODEL and then routing the predicted flows through a simple model that accounted for temporal changes in DOC. Conceptually the DOC model can be taken to represent a terrestrial (soil) reservoir in which DOC builds up during low flow periods and is flushed out when infiltrating meltwaters cause the water table to rise into this "reservoir". Concentrations of DOC measured in the upper soil and in streamflow were compared to model simulations. The simulated DOC response provides a reasonable reproduction of the observed dynamics of DOC in the stream at Deer Creek. Keywords: Carbon; Hydrology; River ecosystems; TOPMODEL
1. I n t r o d u c t i o n A quantitative understanding of the factors controlling the variation of dissolved organic carbon (DOC) in headwater streams is needed for a number of reasons, not the least of which is because D O C interacts strongly with other dissolved substances (heavy metals in particular) and plays an important role in the transport of contaminants. In Deer Creek near Montezuma, Colorado, measurements of dissolved organic carbon (DOC) * Corresponding author.
from July 1991 to August 1992 show rapid decreases in concentration from a peak very early in the snowmelt period (on the ascending limb of the hydrograph); the peak concentration of D O C in the stream occurs well before the peak discharge. A m o n g the main factors controlling D O C variation in such headwater streams are the hydrological catchment responses, including flow paths through and residence times of water in the catchment. Models of the relationships between hydrological flows and D O C transport are important for quantifying the hypothesized mechanisms and thereby indicating whether observations are consistent with hypotheses.
0304-3800/96/$15.00 © 1996 Elsevier Science B.V. All rights reserved SSDI 0304-3800(95)00049-6
184
E. IV.. Boyer et al. / Ecological Modelling 86 (1996) 183-188
We examined possible hydrological mechanisms to explain the observed variability of D O C in D e e r Creek, by first simulating the hydrological response of the catchment using T O P M O D E L and then routing the predicted flows through a simple model that accounted for temporal changes in DOC. Conceptually the D O C model can be taken to represent a terrestrial (soil) reservoir in which D O C builds up during low flow periods and is flushed out when infiltrating meltwaters cause the water table to rise into this "reservoir". To evaluate the model, D O C concentrations were measured in the u p p e r soil and the stream at several locations. Water sampled from lysimeters in the u p p e r soil showed an accumulation of D O C during periods of low flow and a pronounced decline in D O C concentrations during snowmelt. These results, along with the general fidelity with which variations in stream D O C are captured using the model, support the hypothesized flushing mechanism.
to weekly later in the summer, for determination of D O C concentration. 3.2. Hydrological modeling
We applied the hydrological model T O P M O D E L (Beven and Kirkby, 1979) to simulate the catchment hydrology. The features of the U V A - U S G S version of the model that we use have been described by Wolock (1993). The topographic index which is at the core of TOPM O D E L , l o g e ( a / t a n f l ) , was calculated from a) Low flow conditions Litter l a y e r \ Upper soil horizon\ ~ Lower soil Soil water zone ~
~
The site of the study is the D e e r Creek catchment, located in Summit County, Colorado near Montezuma. The catchment is mountainous, ranging in elevation from about 3350 m to 4120 m, and drains an area of 10.6 km 2. Approximately half of the catchment is above the tree line.
3. M e t h o d s
~
..y,~'~~
Water
.... , T °\Wotor " °'tob,o
......
= s,2= .
2. S i t e d e s c r i p t i o n
,
Aquifer Groundwater
~~G1.~:~et:rmovernent
zone
b) High flow conditions Litter l a y e r ~ Upper soil horizon~ __.'~_,, Lowereoll horizon,~ ~ Water table r i d g e ~
~ ~
~"
~ -~
Water
3.1. Field instrumentation
Details of the field m e a s u r e m e n t procedures are given by Boyer (1993). Briefly, a stream gage was installed at the mouth of the catchment that provided a stage record from which the daily discharge was computed. Tension lysimeters, which provide water samples from the u p p e r soil, were installed at several locations on hillslope transects. W a t e r samples from the stream and soil lysimeters were collected, at a frequency ranging from daily near the beginning of snowmelt
" ' ' ' ' ' " ~ ~
A q u i f • r
'~"~---. Water rnovement
Groundwater zone
"Gravel in strearnbed
Fig. 1. Hydrological flushing of soil water. (a) During periods of low flow, subsurface flow is through lower soil horizons. (b) During high-flowconditions (spring snowmelt), the water table rises and flow to the stream is through the upper soil horizons as well. (Taken from Hornberger et al., 1994.)
E. W. Boyer et al. / Ecological Modelling 86 (1996) 183-188
D E M data as described by Wolock (1993). Average soil depth was assumed to be 1 m; an " u p p e r horizon", the presumed reservoir for high D O C levels, was assumed to be 0.5 m thick. TOPM O D E L was calibrated for Deer Creek using the Rosenbrock optimization algorithm. (See Hornberger et al., 1985, for a description of the use of the Rosenbrock method in conjunction with T O P M O D E L . ) The parameters optimized were two soil parameters, one intended to represent the rate of decrease of hydraulic conductivity with depth divided by porosity and the other the hydraulic conductivity of the surface soil (see Beven and Wood, 1983), and two snowmelt parameters, one the threshold temperature for melting and the other the coefficient relating the maximum rate of melting to the average daily temperature (see Bras, 1990).
185
snowmelt (Fig. 1). This would suggest that D O C in the vadose zone might build up during periods of low flow (through leaching from overlying organic-rich litter and through microbial activity) and that this D O C would be pushed out into the stream during periods of high flow. Such a conceptual model would indicate that concentrations of water flowing from the soil to the stream should be high in the initial part of a rainstorm or snowmelt event and then decrease as meltwaters continue to flush the zone formerly above the water table but now exposed to significant streamward flow of water. The simplest mathematical representation of flushing such as envisioned in the conceptual model is from a continuously stirred tank reactor (CSTR). In response to the influx of a flushing solution, such a simple device produces an exponential decline in concentration in the outflow solution with time. We constructed a simple mixing model, based on a CSTR, to explain temporal variation of D O C concentrations in the Snake River. We envision two subsurface reservoirs: one representing the " u p p e r soil" and one representing the "lower
3.3. Chemical modeling The conceptual model is that D O C in riparian and hillslope soils is "flushed out" by the formation of a ground-water ridge during periods of
Simulation o f DOC in Deer Creek
o
O0
~_3
o
e~ 2
•
~o
0 11-Jul-91
09-Sep-91
08-Nov-91 - -
07-Jan-92
simulated DOC
0"/-Mar-92 06-May-92 e
05-Jul-92
observed DOC
Fig. 2. Simulatcd and observed values of D O C for Deer Creek.
03-Sep-92
E. W. Boyer et al. / Ecological Modelling 86 (1996) 183-188
186
soil". Flows through the two subsurface reservoirs are apportioned on the basis of results from T O P M O D E L . We use a simple proration to apportion flow to the u p p e r reservoir. We assume that D O C in the lower reservoir is constant. From data we estimate that the D O C in the lower reservoir is 1.2 mg 1-1. To account for build-up of D O C in the soil, we use the model of Grieve (1991). D O C in the u p p e r reservoir varies according to: DOCupper(t
D O C s t r e a m = (FlOWoverland * D O C s n o w + FIOWuppe r * D O C u p p e r
+ Fl°wl . . . . * D O C I . . . . )/FlOWtotal
4. Results
+ 1)
= D O C u p p e r ( t ) + a * 10 0"04T
- (1 - e - k r ) * DOCupper(t ) -
The D O C in the stream is calculated as a simple mixture of waters from direct snowmelt and from the upper and lower soil reservoirs:
(1 - qupper( t ) / Vupper( t ) ) * D O C upper( t )
where t indicates time, T is temperature, a and k are rate constants for evolution and decay of D O C (see Grieve, 1991), qupper and Vupper are, respectively, the flow through and wetted volume of the u p p e r soil reservoir, and the time step is one day. Although the appropriate t e m p e r a t u r e is a soil temperature, we used air t e m p e r a t u r e as a surrogate. We used a = 0.11 and k = 0.002.
The simulated hydrological response for 19911992 was used as input to our simple D O C model. The simulated D O C response provides a reasonable reproduction of the observed dynamics of D O C in the stream at D e e r Creek (Fig. 2). This was expected from the results presented by Hornberger et al. (1994) who successfully applied this model to describe D O C dynamics in the Snake River, a stream that is confluent with D e e r Creek immediately downstream of our study reach. We attribute the apparent "delay" in the simulated D O C response to the fact that the initial time of rise is late in the simulated discharge hydrograph;
Simulation o f D O C in upper soil 35
30
25
2o r~ 15
10 f I l-Jul-91
09-Sep-91
08-Nov-91
07-Jan-92
07-Mar-92
06-May-92
05-Jui-92
Fig. 3. Simulated values of DOC in the upper soil reservoir.
03-Sep-92
E.W. Boyer et al. /Ecological Modelling 86 (1996) 183-188
187
5. Conclusions
presumably, the t e m p e r a t u r e index model calculates the initiation of significant snowmelt to be later than when the melt is actually initiated. The model captures the essence of the sharp decline in D O C concentrations in the stream after a peak early in the melt hydrograph. The real test of the modeling approach is whether the simulated D O C in the hypothetical soil reservoir was in any sense "close to" that measured in soil solution. Simulated D O C in the u p p e r soil reservoir ranged from about 5 mg 1-1 to about 35 mg 1-1 with p e a k concentrations preceding the initiation of the snowmelt runoff (Fig. 3). Peak measured concentrations of D O C in soil lysimeters ranged up to more than 80 mg 1-1 but most showed peak concentrations in the 20 to 50 mg 1-~ range (Boyer, 1993). Results from a riparian lysimeter, which may be most appropriate for conditions that we are simulating, show D O C concentrations that are very similar to those simulated (Fig. 4).
Neither the simulated discharges from T O P M O D E L nor the simulated D O C values from the C S T R model can be considered to be even nearly perfect representations of the data. Nevertheless, we conclude that the model results are broadly consistent with the available data collected in the stream and in the upper soil of the D e e r Creek catchment. Given the available data, we cannot reject the simple model as a valid description of processes occurring in the D e e r Creek system. The m e a s u r e m e n t s of soil D O C from the field sampling program in 1991 and 1992 lend credence to the model over and above that conferred from the simulation of the stream dynamics. As is often the case, examination of existing data in light of the modeling results suggests additional work that is necessary to more rigor-
D O C in an upper soil lysimeter 50
0.8
40
0.6
~30 o
20
10
0.4 .~
/
0.2
0 06-Apr-92
0 06-May-92
05-Jun-92 o
05-Jul-92 DOC
- -
04-Aug-92
03-Sep-92
03-Oct-92
discharge
Fig. 4. Observed concentrations of D O C in a shallow riparian soil. Stream discharge is shown to indicate the timing of the snowmelt event.
188
E.W.. Boyer et al. /Ecological Modelling 86 (1996) 183-188
ously test t h e h y p o t h e s e s e m b e d d e d in th e m o d e l a nd to r e f i n e t h e m o d e l itself so that c o m p a r i s o n s b e t w e e n m o d e l results and o b s e r v a t i o n s are rend e r e d less a m b i g u o u s . In particular, th e spatial a nd t e m p o r a l v a r i a t i o n s o f s n o w m e l t a n d soil p r o p e r t i e s , in a d d i t i o n to t o p o g r a p h y , should be c o n s i d e r e d in t h e r e f i n e m e n t o f th e c o n c e p t u a l model.
Acknowledgements This w o r k was s u p p o r t e d in p a r t by t h e N a tional S c i e n c e F o u n d a t i o n t h r o u g h G r a n t E A R 9304794.
References Beven, K.J. and Kirkby, M.J., 1979. A physically based, variable contributing area model of basin hydrology. Hydrol. Sci. Bull., 24: 43-69.
Beven, K.J. and Wood, E.F., 1983. Catchment geomorphology and the dynamics of runoff contributing areas. J. Hydrol., 65: 139-158. Boyer, E.W., 1993. Hydrology and the variation of dissolved organic carbon in soil and stream waters of two headwater catchments; Summit County, Colorado. M.S. Thesis, Department of Environmental Sciences, University of Virginia. Bras, R.L., 1990. Hydrology: an Introduction to Hydrologic Science. Addison-Wesley, Reading, MA, 643 pp. Grieve, I.C., 1991. A model of dissolved organic carbon concentrations in soil and stream waters. Hydrol. Processes, 5: 301-307. Hornberger, G.M., Beven, K.J., Cosby, B.J. and Sappington, D.E., 1985. Shenandoah watershed study: calibration of a topography-based, variable contributing area hydrological model to a small forested catchment. Water Resour. Res., 21: 1841-1850. Hornberger, G.M., Bencala, K.E. and McKnight, D.M., 1994. Hydrological controls on dissolved organic carbon during snowmelt in the Snake River near Montezuma, Colorado. Biogeochemistry, 25: 147-165. Wolock, D.M., 1993. Simulating the variable-source-area concept of streamflow generation with the watershed model TOPMODEL. U.S. Geological Survey Water-Resources Investigations Report 93-4124.