Estimation of annual runoff using supraglacial channel geometry derived from UAV surveys of Qiyi Glacier, northern Tibetan Plateau

Abstract. Due to difficulties in direct field observation and uncertainties in glacier runoff models, accurately estimating the glacier runoff remains one of the foremost challenges in cryospheric science. Using a digital elevation model (DEM) and orthophotos (both with a resolution of 5 cm) obtained from an unmanned aerial vehicle (UAV), this study developed a novel remote sensing method for estimating the annual discharge of the supraglacial channel over Qiyi Glacier in the northern Tibetan Plateau, which contributes to the majority of the glacier runoff. Our results showed that the catchment areas of the six main supraglacial channels covered 92.02 % of the total glacier area and transported 89.43 % of the annual surface meltwater yield (each ranging from 0.07×106 to 0.66×106 m3). Some geometric parameters of the supraglacial channels (including lateral deviation, gradient, and width) were selected to predict the annual discharge using a stepwise regression model, which explained ∼78.2 % of the variance in the measurement-based glacier annual discharge, with the explained variance increasing to 81.8 % after five-point moving average filtering. In comparison, a nonlinear regression model incorporating only the lateral deviation and specific gradient, which were more easily obtained practically, performed somewhat less well, accounting for 66.2 % of the discharge variation; however, the explained variance increased to 81.4 % after five-point filtering. If satellite remote sensing data with meter-level spatial resolution are available for a specific glacier research area, our regression models, based solely on the UAV-derived supraglacial channel network, will be a promising solution for monitoring changes in annual glacier discharge.