Advancing Weather Forecasting, Seasonal Climate Predictions, and Agricultural Yield Forecasting Using dynamical modelling and machine learning

This PhD project focuses on improving weather and climate forecasting across multiple time scales, ranging from short-range wind prediction to subseasonal and seasonal (S2S) climate forecasting. The research integrates dynamical models with machine learning techniques to enhance forecast accuracy, reliability, and agricultural applicability. At the short-range scale, the study investigates wind prediction over narrow and complex terrain using the Weather Research and Forecasting (WRF) model combined with Artificial Neural Networks (ANNs). Machine learning is applied to correct systematic model biases and improve wind speed and direction forecasts in regions where topography strongly influences atmospheric flow. At subseasonal and seasonal time scales, the research develops probabilistic rainfall forecasts using both dynamical seasonal prediction systems and machine learning methods. Particular emphasis is placed on improving the reliability of probabilistic seasonal climate predictions and the prediction of seasonal rainfall onset, which is critical for agricultural planning in Africa. Machine learning algorithms are used to better capture nonlinear climate teleconnections and reduce forecast overconfidence. In addition, the project examines the impact of dynamical downscaling of global seasonal forecasts on maize production. Seasonal climate forecasts are integrated into crop models such as WOFOST to assess how improved rainfall predictions influence simulated maize yield. This allows evaluation of the real-world agricultural value of enhanced climate forecasts. Overall, the research aims to bridge atmospheric science and agricultural applications by delivering improved multi-scale forecasting systems. The outcomes are expected to support farmers through better-informed planting decisions, improved risk management, and enhanced food security across climate-sensitive regions of Africa.