Evaluation of a climate simulation over the Yellow River Basin based on a regional climate model (REMO) within the CORDEX

Understanding the current regional climate is of great importance for better estimating future climate change and assessing its influence on water resources and ecosystems. The Yellow River provides important water resources for industrial and domestic activities. However, climate simulations over the Yellow River Basin (YRB) have rarely been attempted. In this study, downscaled mean climate over the upper and middle reaches of the YRB using the REgional MOdel (REMO) at 0.22° spatial resolution under the Coordinated Regional Downscaling Experiment (CORDEX), driven by ERA-Interim reanalysis, was evaluated by comparison against one gridded observational dataset (CN05.1) for the period 1982–2016. Two reanalysis datasets (ERA5 and ERA-Interim) were used to reveal possible causes for the differences between the simulations and observations. REMO reproduced the spatial patterns of mean temperature satisfactorily despite some evident deviations. REMO had a predominantly mixed bias (a cold bias for the upper reach and a warm bias for the rest) in all four seasons. Due to the effects of differential snow cover, temperature biases were elevation-dependent, especially in winter months. REMO broadly reproduced the spatial variability of precipitation with a wet bias, which could be largely attributed to anomalous moisture flux transport. In terms of interannual variability, the observational data showed significant warming across all four seasons, while REMO simulated weaker warming trends. Precipitation trends were positive in all seasons except in summer, but REMO failed to capture the trends in winter and spring. The observed elevation-dependent warming (EDW) was reproduced by REMO, except in spring. The EDW was likely to be explained by the snow-albedo feedback, owing to the apparent decrease in snow cover at high elevations.