A radiation-derived temperature-index snow routine for the GSSHA hydrologic model

Accurate estimation of snowpack is vital in many parts of the world for both water management and flood prediction. Temperature-index (TI) snowmelt models are commonly used for this purpose due to their simplicity and low data requirements. Although TI models work well within lumped watershed models, their reliance on air temperature (and potentially an assumed lapse rate) as the only external driver of snowmelt limits their ability to accurately simulate the spatial distribution of snowpack and thus the timing of snowmelt. This limitation significantly reduces the utility of the TI approach in distributed hydrologic models because spatial variability within the watershed, including snowpack and snowmelt, is usually the primary reason for selecting a distributed model. In this paper, a new radiation-derived temperature index (RTI) approach is presented that uses a spatially-varying proxy temperature in place of air temperature within the TI model of the fully-distributed Gridded Surface Subsurface Hydrologic Analysis (GSSHA) watershed model. The RTI is derived from a radiation balance and includes spatial heterogeneity in both shortwave and longwave radiation. Thus, the RTI accounts for more local variation in the available energy than air temperature alone. The RTI model in GSSHA is tested at the Senator Beck basin in southwestern Colorado where observations for snow water equivalent (SWE) and LandSat-derived images of snow cover area (SCA) are available. The TI and RTI approaches produce similar SWE estimates at two non-forested and relatively flat sites with SWE observations. However, the two models can produce very different SWE values at sites with forests or topographic slopes, which leads to significant differences in the basin-wide SWE values of the two models. Furthermore, the RTI model provides better basin-wide SCA estimates than the TI model in 75% of the LandSat images analyzed.