Evaluation of the ERA5‐Land Reanalysis Data Set for Process‐Based River Temperature Modeling Over Data Sparse and Topographically Complex Regions

Abstract

AbstractModels developed to capture underlying river processes over long historical periods and varying hydrologic conditions provide confidence for subsequent forecasting applications. However, many areas lack the weather data needed to develop process‐based models over these long periods. Climate reanalysis data sets (CRDs) are increasingly used as surrogates for historical meteorology, but their use in river temperature models is still relatively new and untested. Testing of temperature models using CRDs in rivers experiencing a range of instream flow, weather, and topographic conditions is needed to validate the application of these data sets. Focusing on the ERA5‐Land CRD, correction methods that relate weather variables and elevation were tested using weather stations surrounding and adjacent to the Colorado River in Grand Canyon. Our findings show that elevation corrections improved air temperature and relative humidity, but negatively impacted wind speed estimates. Two‐year river temperature model simulations in a 387‐km segment of the Colorado River in Grand Canyon and a 576‐km segment of the Green River showed that using elevation‐corrected ERA5‐Land inputs produced lower mean errors at downstream river locations when compared to predictions using elevation‐corrected ground‐based inputs. Better river temperature predictions when using ERA5‐Land are attributed to the ability to represent spatial variability in weather conditions over these large areas. These promising results persisted when spatially coarsened ERA5‐Land inputs were used. This study highlights the importance of having spatially varying weather information, even at relatively coarse resolutions, when modeling physical processes over large spatial scales and suggests confidence in using CRDs for obtaining this information.