Author(s): Lamboni Batablinle; Lawson Latevi; Agnide Emmanuel Lawin; Mani Kongnine Damegou; Kolani Lankodjoa

Linked Author(s):

Keywords: Rtificial intelligence; Machine learning; LSTM; Precipitation forecasting; Temperature forecasting; Flood risk; Togo; West Afric

Abstract: Rainfall and temperature are pivotal elements of the hydrological cycle, exerting a profound influence on ecosystems, agriculture, and human societies. Achieving accurate and timely orecasting of precipitation and temperature events is crucial to mitigating social, environmental, and economic losses, particularly in regions highly susceptible to climate variability, such as Togo in West Africa. Traditional forecasting techniques, which frequently rely on numerical weather prediction models, often fall short in capturing the nonlinear dynamics of climatic systems due to the complexity of the equations involved. This study seeks to develop a refined model for predicting rainfall and temperature using advanced machine learning (ML) approaches, with a primary emphasis on long short-term memory (LSTM) networks to enhance predictive precision. Models such as convolutional neural networks (CNNs), gated recurrent units (GRUs), and LSTM were applied to a comprehensive dataset of historical rainfall and temperature records from multiple meteorological stations across Togo. Among the models tested, the LSTM network demonstrated superior performance, achieving accuracy rates of 97.27% and 93.81% for rainfall and temperature predictions, respectively. The model’s robustness in handling data irregularities and detecting essential patterns further bolstered the reliability of forecasts. These findings present significant potential for bolstering disaster preparedness and climate resilience efforts in Togo, where extreme weather events such as floods and droughts are recurring challenges. This research underscores the efficacy of LSTM-based ML algorithms in forecasting extreme weather conditions, offering vital tools to strengthen early warning systems and climate resilience across West Africa.

DOI: https://doi.org/10.64697/978-90-835589-7-4_41WC-P1840-cd

Year: 2025