P19 Adjoint sensitivity analysis of FARMS for Forecasting Variables of WRF-Solar.
Yang, Jaemo, Manajit Sengupta, Yu Xie, National Renewable Energy Laboratory, Pedro A. Jimenez, and Ju-Hye Kim, National Center for Atmospheric Research
ABSTRACT: This study presents the development and application of an adjoint model for investigating the sensitivity of solar radiation to forecasted variables from WRF-Solar. The first part of this study focuses on developing an adjoint model for the Fast All-sky Model for Solar Applications (FARMS) to investigate the input variables having the highest sensitivity to global horizontal irradiance (GHI), direct normal irradiance (DNI), and diffuse horizontal irradiance (DHI), which are the output variables. The applicability and usefulness of the adjoint sensitivity approach are demonstrated by conducting a sensitivity analysis under various scenarios defined by leveled values for the input variables. Real physical data is obtained from the National Solar Radiation Data Base (NSRDB; 4-km horizontal resolution for each 30-min interval) and processed to calculate the leveled FARMS inputs (i.e. lower/upper quantiles and median). This preliminary study uses elasticity values to understand the sensitivity of solar radiation to the input variables (e.g. aerosol optical depth, total precipitable water, and cloud optical depth) of FARMS. This presentation will illustrate the implemented methodology and the obtained sensitivity results for FARMS, as well as future research steps that will lead to the development of high-quality probabilistic solar forecasts.