A 24.5-Year Global Dataset of Direct Normal Irradiance: Result from the Application of a Global-to-Beam Model to the NASA GEWEX SRB Global Horizontal Irradiance

Abstract:

The DIRINDEX model has previously been applied to the NASA GEWEX SRB Release 3.0 global horizontal irradiances (GHIs) to derive 3-hourly, daily and monthly mean direct normal irradiances (DNIs) for the period from 2000 to 2005 (http://dx.doi.org/10.1016/j.solener.2014.09.006), though the model was originally designed to estimate hourly DNIs from hourly GHIs. Input to the DIRINDEX model comprised 1.) the 3-hourly all-sky and clear-sky GHIs from the GEWEX SRB dataset; 2.) the surface pressure and the atmospheric column water vapor from the GEOS4 dataset; and 3.) daily mean aerosol optical depth at 700 nm derived from the daily mean aerosol data from the Model of Atmospheric Transport and CHemistry (MATCH). The GEWEX SRB data is spatially available on a quasi-equal-area global grid system consisting of 44016 boxes ranging from 1 degree latitude by 1 degree longitude around the Equator to 1 degree latitude by 120 degree longitude next to the poles. The derived DNIs were on the same grid system. Due to the limited availability of the MATCH aerosol data, the model was applied to the years from 2000 to 2005 only. The results were compared with ground-based measurements from 39 sites of the Baseline Surface Radiation Network (BSRN). The comparison statistics show that the results were in better agreement with their BSRN counterparts than the current Surface meteorology and Solar Energy (SSE) Release 6.0 data (https://eosweb.larc.nasa.gov/sse/). In this paper, we present results from the model over the entire time span of the GEWEX SRB Release 3.0 data (July 1983 to December2007) in which the MERRA atmospheric data were substituted for the GEOS4 data, and the Max-Planck Aerosol Climatology Version 1 (MAC-v1) data for the MATCH data. As a consequence, we derived a 24.5-year DNI dataset of global coverage continuous from July 1983 to December 2007. Comparisons with the BSRN data show that the results are comparable in quality with that from the earlier application.