ARTIFICIAL INTELLIGENCE APPLICATIONS IN CLIMATE CHANGE MONITORING AND ENVIRONMENTAL RISK PREDICTION IN SOUTHWEST NIGERIA

Abstract

Climate change poses escalating threats to ecosystems, agriculture, water resources and human
livelihoods, with rising temperatures, erratic rainfall and more frequent extreme events demanding
better tools for monitoring and anticipating environmental risk. This study developed, validated
and evaluated an artificial intelligence-based system for climate change monitoring and
environmental risk prediction in Southwest Nigeria. A design-and-development research approach
was employed, combining environmental sensing, machine-learning model development and field
validation across six purposively selected sites in Southwest Nigeria. Environmental sensor nodes
and automatic weather stations were deployed to capture temperature, humidity, rainfall, wind
speed and solar radiation data, combined with historical climate and satellite datasets. Machine
learning models (Random Forest, Gradient Boosting, Long Short-Term Memory) were trained for
climate monitoring and flood risk and heat stress prediction. The LSTM model achieved the best
performance for temperature prediction (RMSE = 0.84°C, MAPE = 3.2%), flood risk prediction
(AUC = 0.91, F1-score = 0.87), and heat stress prediction (AUC = 0.89, F1-score = 0.84). Model
performance was validated against observed outcomes, demonstrating that AI-based prediction is
feasible under regional data conditions with appropriate model selection. Expert validation rated
the system highly for functionality (3.60), predictive accuracy (3.40), and usability (3.60). The
developed decision-support dashboard provides real-time monitoring and forecasting of
environmental risks. The implementation framework provides a practical roadmap for climate
monitoring and early warning in Southwest Nigeria. The findings demonstrate that AI-based
climate monitoring and risk prediction is technically feasible and practically valuable in the
Nigerian context, and they provide a replicable tool to support climate resilience and early warning
in the region.