Enhancing Solar Radiation data using High-resolution GOES Cloud Products

The National Solar Radiation Database (NSRDB) is a comprehensive dataset offering high-temporal and spatial resolution solar radiation and meteorological data. The present version of NSRDB is constructed via modeling techniques utilizing multi-channel measurements obtained from the Geostationary Operational Environmental Satellite (GOES). This yields solar radiation at a 2 km resolution in every 5-minute interval. With the 500 m resolution visible bands of GOES-16 and GOES-17, cloud products can now be obtained at a finer resolution. This research aims to explore consequential enhancements brought by the integration of this refined cloud data into the NSRDB. One year cloud products from Goodwin Greek, MS; Sioux Falls, SD; Penn State, PA; and Table Mountain, CO are used in this study. The 1-min resolution surface solar radiation observation from the Surface Radiation Budget (SURFRAD) network is averaged over 5-minute intervals to align with the NSRDB timestamps. The latest NSRDB algorithm is used to compute high-resolution solar radiation in 2020 for each 500 m satellite pixel. The computation is averaged across the neighboring pixels surrounding the surface sites. To streamline computational demand, we also estimate cloud fraction using data from the 169 satellite pixels surrounding the ground station. The all-sky global horizontal irradiance (GHI) and direct normal irradiance (DNI) are derived using the cloud fraction and the computed clear-sky and cloudy-sky radiation. Our preliminary results suggest that the integration of the 500 m data contributes to the reduction of mean bias error and mean absolute error for both GHI and DNI, in contrast to the conventional NSRDB algorithm. Furthermore, when compared to the average of solar radiation across satellite pixels, the incorporation of cloud fraction demonstrates a supplementary reduction in solar radiation bias.