Estimation d'irradiations solaire globales horaires à partir des images MSG par ANN

Abstract

Images from the Meteosat MSG satellite are a rich source of data for scientific research. The daily availability of these data sources makes them an indispensable tool for monitoring, analyzing and understanding all types of weather phenomena. It is the observation of the Earth and the collection of data and knowledge about it The objective of our research is to determine the amount of solar radiation in Algeria. To achieve this goal, we used the GisTeL model, which uses satellite imagery in the visible channel. This model is based on determining the sky index in three different situations: clear, partially overcast, and completely overcast. We also used Keras's detailed learning method and chose linear regression as the method for estimating global horizontal solar radiation. This choice was justified by the simplicity and efficiency of the method for determining the images, which allowed us to calculate the global solar radiation on the surface of the Earth We proposed to use the linear regression method to improve the quality of the results obtained. Estimated values of global solar radiation are compared to measured values to validate our results. We used the UREAR weather station in Ghardaia as an example. The results obtained demonstrate the validity of the model used, with good margins of error.