Evaluating the Snow Module of the LISFLOOD Model with Remotely Sensed Snow Cover

LISFLOOD is a comprehensive large-scale operational hydrological model widely used in Europe to simulate diverse hydrological processes, including snowmelt, which is handled through a degree-day-based snow module (Van Der Knijff et al., 2010). The snowmelt coefficient in this module is traditionally calibrated against discharge data. This study evaluates the performance of LISFLOOD’s current snow module and explores an alternative calibration approach based on snow cover area (SCA) observations. Nine hydrological basins across Europe located in Italy, Switzerland, Austria, Germany, France, Spain, Slovakia, and Sweden were selected for this analysis. They represent a range of climatic and morphological characteristics, from mountainous regions such as the Alps and Pyrenees to the flatter terrains of Scandinavia. Their strong snow influence, with persistent snow cover for significant portions of the year, makes them ideal for assessing snow processes. 

First, we evaluated several operational satellite-based snow cover products. This included an intercomparison of data gaps and agreements, benchmarked against a novel product that integrates Sentinel-2 and MODIS datasets using gap-filling and downscaling techniques to achieve high temporal and spatial resolution (Premier et al., 2021). Next, the snowmelt coefficient was estimated on a pixel-wise basis by fitting the modeled snow cover fraction (SCF) -derived from snow water equivalent (SWE) in LISFLOOD - with observed satellite-based SCF. This involved an appropriate parametrization to convert SWE to SCF and an optimization routine to minimize errors between modeled and observed SCF. The resulting spatially distributed snowmelt coefficient represents a novelty compared to the current LISFLOOD setup, where coefficients are uniform across subcatchments. 

Our findings show that LISFLOOD’s current configuration performs well when validated against independent satellite-based snow cover products. While the newly optimized snowmelt coefficients differ considerably from previously calibrated values, they do not introduce significant changes in terms of simulated discharge. However, notable effects are observed in the timing and magnitude of SWE and snowmelt processes, underscoring the potential for improved representation of snow dynamics in LISFLOOD. 

 

References

Premier, V., Marin, C., Steger, S., Notarnicola, C., & Bruzzone, L. (2021). A novel approach based on a hierarchical multiresolution analysis of optical time series to reconstruct the daily high-resolution snow cover area. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 14, 9223-9240.

Van Der Knijff, J. M., Younis, J., & De Roo, A. P. J. (2010). LISFLOOD: a GIS‐based distributed model for river basin scale water balance and flood simulation. International Journal of Geographical Information Science, 24(2), 189-212.