Evaluation of Radiative Flux Datasets over the Great Lakes Region for Climate Model Validation Applications

Radiative flux patterns within the Great Lakes system are a crucial driving force for the hydrological and meteorological processes that determine the region’s unique climate. Specifically, shortwave (solar) radiation absorbed by the surface strongly influences the extent of lake ice cover, thereby modulating seasonal temperatures and precipitation patterns. To gain a more comprehensive understanding of climate variability and change within the Great Lakes region, Numerical Weather Prediction (NWP) models are strategically employed. However, surface radiative fluxes must be accurately simulated to effectively model things such as lake ice cover, lake-effect snow, and regional temperature climatology. Thus, performing validations on modeled surface radiative fluxes within the Great Lakes region can help resolve biases for these meteorological processes. This study aims to identify a reliable gridded dataset that can be utilized to conduct these kinds of spatiotemporally comprehensive model validations. To accomplish this, two reanalyses (ERA5 & MERRA2) and two satellite-based (CERES-EBAF & CLARA-v2) datasets of monthly mean surface downwelling shortwave and longwave fluxes (2001-2020) are compared against a robust collection of in-situ measurements. Lake and land radiative flux observations are sourced from the Great Lakes Evaporation Network (GLEN), Fluxnet/Ameriflux stations, and NDBC Buoys. While past studies have conducted similar evaluations (e.g. Moukomla & Blanken, 2016 and White et al. 2011), there is very limited research on the performance of current radiative flux datasets over the Great Lakes region. Preliminary results indicate that satellite-based datasets, CERES-EBAF and CLARA-v2, outperform reanalysis datasets for almost all in-situ locations considered. Average RMSE values for CERES-EBAF and CLARA-v2 were 17.58 W/m² and 19.39 W/m² respectively, while average RMSE values for ERA5 and MERRA2 were 21.03 W/m² and 25.7 W/m² respectively. Interestingly, all datasets were able to more accurately capture radiative flux magnitudes over land than over the lakes, with RMSE values ~30% lower over land. All datasets reasonably captured the seasonal trends in surface radiative fluxes, with average correlation coefficients for all datasets falling above 0.9 (R > 0.9). These initial findings illustrate the overall capacity of these four datasets to serve as the basis for NWP model validations of surface radiative fluxes in the Great Lakes region, while highlighting the need to expand surface radiative flux observation networks over the Great Lakes themselves. As of the writing of this abstract, there has been a newly published dataset, CLARA-v3, with extended availability and additional temporal resolutions, which shall be integrated into this analysis.