Skip to Main content Skip to Navigation

Assimilation de données satellitaires géostationnaires dans des modèles atmosphériques à aire limitée pour la prévision du rayonnement solaire en région tropicale

Abstract : The variability of solar irradiance necessitates to limit the instantaneous feed-in of solar power to electricity grids. An improvement of solar irradiance forecasts would allow to increase the defined threshold limits, especially in non-interconnected zones such as Reunion Island. Achieving higher forecast accuracy is particularly challenging in the case of tropical islands due to pronounced convection and local thermal circulations. Limited-area numerical weather prediction (NWP) models allow to forecast cloud processes and solar irradiance at high spatio-temporal resolutions of a few kilometres and minutes. Nevertheless, they often fail to accurately predict cloudiness evolution and thus tend to overestimate solar irradiance. Refining the initial conditions of regional models in terms of clouds is an efficient means for improving short-term cloud cover and irradiance forecasts. The assimilation of geostationary meteorological satellite observations can achieve this improvement. Nevertheless, a variety of satellite data assimilation (DA) approaches exist and research has focused on mid-latitudes so far. This thesis deals with the assimilation of geostationary satellite observations with limited-area models in the southwestern Indian Ocean. In a first step, the state of the art in terms of existing approaches for radiance and cloud property retrieval assimilation with regional-scale models is reviewed. In consequence, one of the most promising approaches is identified and applied to the southwestern Indian Ocean. In the performed experiments, multi-phase cloud water path retrievals from NASA Langley's SatCORPS cloud products are assimilated with an ensemble Kalman filter using the Weather Research and Forecasting model. A 41-member ensemble at 12 km grid spacing is applied with a DA cycling interval of 6 hours. The Data Assimilation Research Testbed and its forward operator for cloud water path are used to assimilate gridded cloud water retrievals in the ice, supercooled liquid, and liquid phase. The impact of this assimilation approach on forecasts of global horizontal irradiance (GHI) is evaluated for austral summer 2017/2018 using pyranometer observations on Reunion Island. A distinct positive impact of the applied method on the first 14 hours of GHI forecasts is found. Different aspects of the forecast improvement due to DA are analysed by means of control experiments without DA, experiments with a nested domain at 4 km grid spacing, and a comparison with operational NWP models. As the utilised gridded cloud products are available globally, the method offers a portable and globally applicable approach that may also be evaluated for other regions of the Earth.
Document type :
Complete list of metadatas

Cited literature [254 references]  Display  Hide  Download
Contributor : Abes Star :  Contact
Submitted on : Sunday, March 1, 2020 - 1:02:35 AM
Last modification on : Friday, July 10, 2020 - 2:12:02 PM
Long-term archiving on: : Sunday, May 31, 2020 - 12:19:38 PM


Version validated by the jury (STAR)


  • HAL Id : tel-02495080, version 1


Frederik Kurzrock. Assimilation de données satellitaires géostationnaires dans des modèles atmosphériques à aire limitée pour la prévision du rayonnement solaire en région tropicale. Géographie. Université de la Réunion, 2019. Français. ⟨NNT : 2019LARE0013⟩. ⟨tel-02495080⟩



Record views


Files downloads