Estimating snow water equivalent using unmanned aerial vehicles for determining snow-melt runoff

Abstract

Although global warming decreases winter snow storage and snow-melt peak discharges in the majority of basins in warm regions, there exist specific catchments in cold regions where climate change increases the risk of snow-melt floods in winters and earlier-than-usual springs. The volume of water contained in a snowpack is quantified by snow water equivalent, the near-real-time estimation of which is essential to issue early warnings against snow-melt floods. We propose a new approach for rapid and high-resolution estimation of snow water equivalent for small mountainous basins, employing unmanned aerial vehicles known as drones. Numerical maps of snow water equivalent are automatically produced for subareas of a river basin by combining drone-based snow depth maps with snow density estimates. The reconstructions for subareas are extrapolated to the entire river basin using a zonal version of multiple linear regression. Validation is conducted on the example of the Kwisa river in southwestern Poland by balancing drone-based snow water equivalent with snow-melt runoff separated using a hydrologic model from discharges continuously measured during a thawing period. Differences between the drone-based and reconstructed snow-melt runoff are found to be of −11.2, +1.6 and +18.9%, providing evidence for the considerable worth skills of the new drone-based flood risk assessment method.