PhD thesis,
Quantifying solar radiation at the earth surface with meteorological and satellite data.
University of Twente Faculty of Geo-Information and Earth Observation (ITC), P.O. Box 217, 7500 AE Enschede, The Netherlands, (January 2014)
Abstract
The energy of the sun drives physical, chemical and biological processes taking place in the earth-atmosphere system, thereby determining the earth's climate and making organic life on earth possible. Solar radiation provides the energy that plants need for growth. The knowledge about the spatial distribution and temporal variation of solar radiation reaching the earth surface is important for crop growth monitoring and yield forecasting. In addition to other weather variables (i.e., air temperature and precipitation), solar radiation is an essential variable required by most crop growth models. Direct measurement of solar radiation is now carried out in most European countries, but the network of measuring stations is too sparse for a reliable interpolation of measured values. Thus, various approaches have been developed to obtain solar radiation estimates for locations where it is not directly measured. These include empirical and physically-based models, numerical weather prediction models, as well as retrievals from satellite observations. This thesis explores these different approaches, and aims at improving the estimation of daily surface solar radiation used for operational crop growth modelling in Europe. The European Commission's MARS Crop Yield Forecasting System is used here as the main application example, as it requires accurate solar radiation data in near real-time to run the crop growth model.
This thesis presents a comprehensive accuracy assessment and rigorous inter-comparison of solar radiation datasets for Europe; these are derived from satellite observations, empirical solar radiation models, and weather prediction models. The thesis moreover provides guidelines on how a long-term seamless gridded time series of daily solar radiation may be constructed for Europe from two currently available products that are derived from geostationary satellites, the European Re-Analysis (ERA-Interim), and weather station data.
Figure: A comparison between daily solar radiation estimates derived from Meteosat First Generation (SIS), Meteosat Second Generation (DSSF), ERA-Interim, and solar radiation estimates interpolated from weather stations (JRC-MARS) demonstrated by the relative root mean square error (RRMSE) and mean bias error (MBE). Positive mean bias indicates that the estimate represented by a row has a higher value than estimate represented by a column.
The two principal findings of the thesis can be summarized as follows:
(1) A 30-year-long time series of daily surface solar radiation can be created by merging a Meteosat First Generation-based dataset (1983-2005, Satellite Application Facility for Climate Monitoring) with a Meteosat Second Generation-based product (since 2005, Land Surface Analysis Satellite Application Facility). These time series can be extended in near real-time by the Meteosat Second Generation-based product. The ERA-Interim reanalysis provided by the European Centre for Medium-Range Weather Forecasts (ECMWF) may be used as a back-up for satellite-based solar radiation estimates for an operational system.
(2) Apart from satellite-based retrievals, solar radiation can be estimated from daily sunshine duration, cloud cover, and air temperature range for every location in Europe using empirical solar radiation models calibrated with one of the two new calibration methods introduced in this thesis. These two calibration methods – auto-calibration and satellite-based-calibration – do not require station measurements of solar radiation, which is employed by the classical model calibration approach. Comparison with station data showed that solar radiation estimated with these newly-calibrated models has an accuracy similar to the retrievals from the models calibrated with measured solar radiation data. Both calibration methods were applied and evaluated only for weather stations in Europe, but the methods should be easily transferable to other continents.
This thesis reveals that Meteosat-derived solar radiation estimates are by far the best source for long-term gridded solar radiation time series for Europe, which can be extended in near real-time. The reliability of such estimates is guaranteed towards the future with the perspective of the forthcoming Meteosat Third Generation, planned to be operational until 2038. Although the thesis focuses on the need of solar radiation estimates for crop growth monitoring, both the methods demonstrated here and the conclusions of this thesis may potentially be valuable for other disciplines, such as in the fields of hydrology, ecology and solar energy.
