Developing an Original Computational Model for Predicting Urban Heat Island Intensity and Its Building Energy Implications in Nigerian Metropolitan Areas

Urban heat island intensity prediction in Nigerian metropolitan areas requires a computational model that integrates urban morphology data, land surface cover characteristics, and meteorological boundary conditions in a physically realistic framework calibrated to tropical West African urban environments, where existing models have not been validated. This study develops and validates an original Urban Heat Island Prediction Model (UHIPM-NG) for Nigerian metropolitan areas through a multi-stage research programme. Comprehensive urban climate monitoring networks were established in three metropolitan areas: Lagos, Kano, and Port Harcourt, comprising 36 monitoring stations per city measuring air temperature, surface temperature, humidity, wind speed, and solar radiation at 15-minute intervals over 24 months. Land use, urban morphology, and surface albedo data were derived from Landsat 8 and Sentinel-2 satellite imagery using original land cover classification algorithms developed for the Nigerian urban context. The UHIPM-NG was developed by modifying the Weather Research and Forecasting model's urban canopy physics with Nigeria-specific surface parameters, calibrated using the monitoring network data. The model's prediction accuracy for mean urban heat island intensity was validated with RMSE of 0.68 degrees Celsius across all three metropolitan areas. The UHIPM-NG was used to conduct original scenario analyses of urban green infrastructure, cool roof deployment, and urban geometry modification as heat island mitigation strategies. Building energy implications of heat island mitigation scenarios were calculated using EnergyPlus for the three metropolitan areas. The model and its parameterisation are made available as open-source tools for Nigerian urban climate research.

Keywords: urban heat island, computational model, Nigerian cities, WRF model, building energy