This paper focuses on the prediction of daily surface solar radiation maps for Reunion Island by a hybrid approach that combines principal component analysis (PCA), wavelet transform analysis and artificial neural network (ANN). The daily surface solar radiation over 18 years (1999-2016) from CM SAF (SARAH-E with 0.05°×0.05° spatial resolution) is first detrended by using the clear sky index. Dimensionality reduction of the detrended dataset is secondly performed through PCA, which results in saving computational time by a factor of eight in comparison to not using PCA. A wavelet transform is thirdly applied onto each of the first 28 principal components (PCs) explaining 95% of the variance. The decomposed 9-wavelet components for each PC are fourthly used as input to an artificial neural network (ANN) model to perform prediction of day ahead surface solar radiation. The predicted decomposed components are finally returned to PCs and clear sky indices, irradiation in the end for re-mapping the surface solar radiation's distribution. It is found that the prediction accuracy is quite satisfying: RMSE is 30.98 W/m2 and the (1 - RMSE_prediction/RMSE_persistence) is 0.409.
The observational data described in this article are collected at several locations in the South-West Indian Ocean (SWIO). Platforms equipped with radiometers and a weather transmitter, and located over Comoros, Madagascar, Mauritius, La Réunion and Seychelles islands, are used to measure incident global and diffuse shortwave radiation and incident global UV A + B-band radiation along with air temperature, relative humidity, wind speed and direction, air pressure and rainfall amount with a sampling frequency of 0.1 Hz. The data are stored as 1-min averages and automatically transmitted to the LE2P-ENERGY lab at the University of La Réunion. The dataset is hosted on the website: https://galilee.univ-reunion.fr, and is uploaded to Zenodo repository. Such a dataset will help in providing information related to solar energy forecasting and assessment for solar energy implementation at a regional and national level in the SWIO.
Solar radiation incident on the Earth's surface is important for the functioning of tropical forests, as it affects the availability of light and water. Due to the lack of in-situ data in tropical forest environments, satellite products and reanalyses are the only ways to estimate solar radiation on a regional scale. An intercomparison of five satellite databases including CERES-EBAF, CERES-SYN1deg, CMSAF-SARAH, CMSAF-CLARA, CAMS-JADE as well as the ERA5 reanalysis, is carried out for the Atlantic coast of Central Africa by evaluating them against two in-situ data sets: the monthly FAOCLIM2 database and original infra-daily data from meteorological stations set up within the framework of ecoclimatic projects. From this inter-comparison we show the differences between these six products and with in-situ data from monthly to daily scales. We also show that the Atlantic coast of Central Africa receives the least amount of solar radiation in all products compared to other regions of Central Africa.
Western Equatorial Africa (WEA) is characterized by a long and cloudy dry season extending from June to September. So far, no study has investigated the intra-seasonal characteristics of this dry season especially its onset and cessation dates. In our study, the onset and cessation dates are determined over the 38-year period 1983–2020, using daily surface solar radiation (SSR) data from CMSAF SARAH-2. The maximum and minimum values of the cumulative anomalies of a regional index, for each year, are used to extract the onset and cessation dates. The mean onset date of the dry season in the region is May 17, the mean cessation date is October 3. We obtain very distinct anomaly patterns of SSR but also of low-level clouds and precipitation before/after the onset/cessation dates. The onset and cessation dates show strong year-to-year variability but no significant trend is detected over the 4 decades studied. Lastly, the cumulative anomalies for each year are also used to classify the dry seasons according to the SSR intra-seasonal evolution. Three types of years are obtained which are associated to different patterns of SST anomalies in the tropics.