Estimating stream chemistry during the snowmelt pulse using a spatially distributed, coupled snowmelt and hydrochemical modeling approach

We used remotely sensed snow cover data and a physically based snowmelt model to estimate the spatial distribution of energy fluxes, snowmelt, snow water equivalent, and snow cover extent over the different land cover types within the Green Lakes Valley, Front Range, Colorado. The spatially explicit snowpack model was coupled to the Alpine Hydrochemical Model (AHM), and estimates of hydrochemistry at the basin outlflow were compared with the baseline AHM approach, which implicitly prescribes snowmelt. The proportions of total meltwater production from soil, talus, and rock subunits were 46, 25, and 29%, respectively, for the baseline simulation without our advanced snowmelt representation. Conversely, simulations in which the AHM was coupled to our distributed snowmelt model ascribed the largest meltwater production to talus (47%) subunits, with 37% ascribed to soil and 16% ascribed to rock. Accounting for these differences in AHM reduced model overestimates of cation concentration during snowmelt; modeled Ca2+ estimates explained 82 and 70% (P values < 0.01) of observations with and without the coupled model, respectively. Similarly, the coupled model explained more variability in nitrate concentrations, with 83 versus 70% (P values < 0.01) explained by the coupled and baseline models, respectively. Early snowmelt over talus subunits was not detected at the basin outflow, confirming earlier reports that deeper flow paths are needed in biogeochemical models of alpine systems. Realistic treatment of snowmelt within these models will allow efforts to improve understanding of flow paths and predict catchment response to increases in atmospheric deposition and climate change.