Using artificial intelligence techniques and standardised precipitation evapotranspiration index for meteorological drought forecasting

Abstract

This thesis investigates the application of Artificial Intelligence (AI) to monitor and forecast drought in the Northeast region of Algeria, utilizing atmospheric circulation indices and the Standardized Precipitation-Evapotranspiration Index (SPEI). The primary objective is to develop accurate forecasting models and identify the atmospheric circulation indices most influential in this region. Drought, a recurrent natural hazard in Algeria, significantly impacts agriculture, water resources, and socio-economic activities. Traditional monitoring methods often fall short in predicting drought onset and severity. This study leverages AI techniques, specifically the Random Forest model, to analyze and interpret large datasets comprising climatic variables and atmospheric circulation indices. The research methodology includes collecting historical climate data and atmospheric circulation indices relevant to the Northeast region of Algeria. The SPEI index, which incorporates both precipitation and temperature data, is used to quantify drought conditions. The Random Forest model is trained and validated to predict SPEI values based on the selected atmospheric indices. Results demonstrate that the Random Forest model achieves high accuracy in forecasting drought, with some atmospheric circulation indices proving to be significant predictors. The findings highlight the potential of AI to enhance drought monitoring systems, offering timely and reliable information for decision-making and resource management. This study not only contributes to the understanding of drought dynamics in Algeria but also provides a framework for integrating AI into environmental monitoring systems. The successful identification of influential atmospheric indices further enriches the scientific knowledge required for developing robust predictive models.