Site-dependent Evaluation of the ECMWF IFS-HRES and IFS-COMPO Intra- and Day-ahead Forecasts with Respect to surface solar irradiances
Résumé
The main focus of the CAMS Radiation Service is to provide high-quality long-term historical irradiance data which is easily integrated in solar resource assessments and other related studies. At the same time, the CAMS system provides a forecast of atmospheric composition, i.e., aerosols, water vapor and ozone. As part of its Integrated Forecasting System (IFS), ECMWF generates the alternative IFS-COMPO forecast which uses additional modules based on the atmospheric composition forecasted by CAMS (Inness et al., 2015). The IFS-COMPO forecast is operated on a spatial resolution of 40 km (~ 3.5°) which is quite coarse compared to the 9 km (~0.1°) of the operational IFS-HRES. At the same time the IFS-COMPO forecasts are based on additional aerosol forecasts as provided by the CAMS service while the operational IFS-HRES runs use an aerosol climatology (Bozzo et al., 2017). Due to their different native spatial resolution and atmospheric modeling, it is crucial to understand how the IFS-COMPO radiation forecast compares to the operational IFS-HRES forecast and whether it could be implemented as an operational product on itself. The present evaluation is done for the full year 2022 on a global scale using ground observations from well-known research solar radiation measurement networks (BSRN, SAURAN, SURFRAD, enerMENA, NREL-MIDC). On a first step all the ground observations are passed thought the expert QC battery of tests developed by the CAMS radiation service team. Then, for the evaluation of the forecasts, the locations of these observations are classified on 5 types: Desert, Polar, Subpixel, Mountainous, Continental. This classification scheme allows to identify strengths and weaknesses and to compare from different perspectives the performances of the forecasts. Following solar energy stakeholders needs, this evaluation is concentrated on the intraday and day-ahead range, that is, the evaluation will use the forecasts up to 48 hours. The forecasts are obtained from the run at 0:00 UTC which ensures that the forecasts are available for all intraday and day ahead market transactions and operational processes related to the energy system stakeholders.
