Snow evolution in a semi-arid mountainous area combining snow modelling and Landsat spectral mixture analysis

Abstract

This study proposes the use of both physically-distributed hydrological modelling in combination with satellite remote sensing images, to study the evolution of the snowpack in the Sierra Nevada mountains, in southern Spain. The snowmelt-accumulation module inside WiMMed (Watershed Integrated Management in Mediterranean Environment) hydrological model was employed, which includes the use of depletion curves to expand the energy and water balance equations over a grid representation. Snow maps obtained from spectral mixture analysis of Landsat images were used to evaluate this model at the study site. The results show a significant agreement between observed and simulated snow pixels in the area, which allows production of sequences of snow maps with greater resolution than the remote sensing images employed. However, some mismatches do appear at the boundaries of the snow area, mainly related to: (a) the great number of mixed pixels; and (b) the influence of the snow transport by wind.