부산대학교 도서관

This study investigates the optimum usage of the new APOLLO Next Generation (APOLLO_NG) cloud retrieval scheme in the Heliosat‑4 solar surface irradiance retrieval scheme. APOLLO_NG replaces the previously used APOLLO scheme and introduces a probabilistic cloud mask instead of binary cloud masking used previously. A cloud probability threshold of 1 % was found optimal for the usage of the Heliosat‑4 cloud scheme. For direct irradiances, the use of an apparent optical depth in optically thin as well as in optically thick, but broken cloud conditions, was investigated. While for thin clouds the cloud optical thickness (COT) modification factor as known from literature was confirmed, a COT modification factor of 0.2 for optically thick, but broken clouds was identified. Furthermore, an improved calibration of the SEVIRI instrument was implemented. The updated method is applied for MSG SEVIRI observations without any additional bias-correction and validated for Europe and Africa against observations from the BSRN and EnerMENA networks. Special focus is placed on the central European region affected mainly by clouds and the North-African region with desert dust aerosols as well as clouds. The hourly SSI showed clear improvements. The relative bias of GHI ranges between −7 and 1.3 % with a mean value of −2.8 %. The relative RMSE of GHI is between 10.3 and 25.5 % and has a mean value of 13.7 %. The relative bias of DIR ranges between −16.6 and 17.8 % and has a mean value of −3.3 %. The relative RMSE of DIR is between 16.7 and 49 % with 29.7 % as mean. For European stations without strong aerosol effects but larger cloud impacts, the average relative GHI and DIR bias is −1.3 % and 3.3 % with mean relative GHI and DIR RMSE of 15.1 % and 30.6 % respectively. The presented method has been implemented into the Copernicus Atmosphere Monitoring Service (CAMS) Radiation Service v4.