Use of satellite snow-cover data for streamflow prediction in the Feather River Basin, California

Abstract

The MODerate Resolution Imaging Spectroradiometer (MODIS)/Terra daily snow-cover 500 m resolution product (MOD10A1) is evaluated through comparisons with ground-snow observations over the western US. The misclassified fractions of MOD10A1 (most common misclassification of snow-covered as non-snow-covered pixels) are generally lower at high elevations than at low elevations, and are greatest during the ablation period and at the beginning of the snow-accumulation period. The MOD10A1 data are used to update the snow-cover status in the Variable Infiltration Capacity (VIC) hydrology model applied to the Feather River Basin, California over the 2000–2008 MODIS period. The effect of MODIS updating is evaluated both on retrospective simulations, and on forecasts made at twice-monthly intervals, both for lead times of 2 weeks, and for seasonal forecasts from winter and spring forecast dates through to the end of July. Evaluations of simulated streamflow using the calibrated standard VIC model and the calibrated VIC with snow-covered area (SCA) updating show that the MODIS updating tends to reduce simulated streamflow and does not necessarily reduce simulation errors. Results for retrospective forecasts showed that assimilation of MODIS data reduced the Mean Absolute Error (E MAE) of 2 week lead-time forecasts during 2000–2008 in the snow-ablation period. However, forecast accuracy for seasonal forecasts during the same period was not improved by assimilation of MODIS data.