Monitoring Satellite-derived Surface Solar Radiation with Near Real Time Reference Data

Abstract:

Geostationary satellite observations of the Earth are increasingly made more frequent. For example, Himawari-8 of Japanese Meteorological Agency takes images of the planet every 10 minutes in multiple bands. Similarly, the GOES-R satellite of the US National Oceanic and Atmospheric Administration (NOAA) will make observations every 5 to 15 minutes. Products, like shortwave (solar) radiation budget at the surface, derived from these observations have or will have similar rapid refresh rates. Routine, near-real time assessment of the quality of these products ideally requires the availability of near-real time reference data. Such near-real time data has recently become available from the NOAA Surface Radiation Budget Network (SURFRAD). These data are disseminated every 15 minutes. However, in contrast to non-real-time data with fully quality control, which have a latency of 24 hours or more, the near-real time data have less quality control applied to them in order to achieve low latency. To assess applicability of this near-real time SURFRAD data for the evaluation satellite products we are using them experimentally to evaluate the quality of Downward Shortwave Radiation at the surface (DSR) retrieved operationally every hour from GOES and made available in the Geostationary Surface and Insolation Product (GSIP) . Metrics (accuracy and precision) are computed to characterize the level of agreement between satellite retrievals and the near-real time reference data. These metrics are then compared with metrics from the evaluation with the non-real time, fully quality controlled reference. The comparison shows that monitoring of DSR with near-real time data is not very different from monitoring it with non-real time data and so DSR retrievals can be evaluated hourly or shorter times depending on reference data availability.