MODIS snow cover data for calibration and evaluation of hydrological models in contrasting environments

This research studies different hydrological models in the catchments with various climates in the USA, and MODIS snow cover maps are introduced to assess and improve the models. In this research, assessment of the value of MODIS snow cover data in the hydrological modeling process is the main objective. Nine MOPEX catchments in the continental US are studied in this research and for each catchment three types of models are used: (1) a basic, lumped 3-box version of the HBV model, (2) a lumped version with snow reservoir, and (3) a semi-distributed formulation of HBV with snow reservoir. MODIS snow cover maps are used for assessing and improving the model snow component. In the models with snow reservoir, snow water storage is reduced to its binary information content: snow/no snow (i.e. 1 or 0), which is then directly linked with spatial pattern in MODIS snow cover data without the need for further assumptions, such as snow densities; the coincidence of snow water storage in the model and observed MODIS snow cover is used as additional calibration/evaluation metric to condition and test the realism of the respective models. In this research, daily meteorological and hydrological data (precipitation, evaporation, discharge, temperature) and maximum eight-day MODIS snow cover maps for the study catchments are used. In calibration, in each time step, the modelled maximum binary 8-day snow reservoir value is compared with MODIS snow cover data for the respective model or its sub-units (in the semi-distributed formulation). Model 03 is semi-distributed, by discretizing each study catchment into individual 200m elevation bands. The results show a large difference of model performance in the catchments with various climates. It is shown that MODIS snow cover data can be valuable to assess the degree to which spatial snow accumulation and ablation dynamics are represented in models, thereby increasing model realism, while not necessarily improving the calibrated performance of models.Civil Engineering and GeosciencesWater ManagementCIE5050-0