Solar resource modeling for CSP: Current status of aerosol-related errors in South-Eastern Asia

The solar resource assessment in areas of interest for CSP has to rely at some point on modeled solar irradiance calculated from atmospheric composition data because on-site ground observations, if any, are limited. In the absence of clouds, regular situation in areas of high solar resource, the critical atmospheric constituent is aerosol, which may undergo sudden changes from day to day and region to region. Atmospheric chemistry models are a convenient source of aerosol information to evaluate solar irradiance. However, their performance varies widely depending on the regional weather characteristics and aerosol sources (e.g., desert dust vs. urban pollution). This work evaluates the reliability of modeled solar irradiance in south-eastern Asia under cloudless conditions when aerosol data are gathered from the CAMS and MERRA2 global atmospheric chemistry models, two of the most advanced models of its class. To minimize the impact of errors from the solar radiation model itself, the high-quality SMARTS spectral solar irradiance model is used. The study is conducted throughout four sites of the BSRN radiometric network over 1 year of observations at each one. Overall, prediction uncertainty of DNI is about three times higher than the prediction uncertainty of GHI.