Examining the Research Gap in Multi-Modal Data Fusion for Urban Climate Monitoring in Nigerian Cities

Urban climate monitoring in Nigerian cities requires integrating data from multiple heterogeneous sources including ground sensors, satellite imagery, and crowd-sourced reports, yet multi-modal data fusion approaches for this purpose have not been evaluated in Nigerian urban contexts. This study examined the research gap in multi-modal data fusion for urban climate monitoring in Lagos and Kano through systematic review and original experimental analysis. A systematic scoping review found that 89 percent of urban climate monitoring fusion studies were conducted in North American, European, or East Asian cities, with zero studies focused on West African urban contexts. Original experimental analysis fused data from 12 weather station sensors, Landsat 8 land surface temperature imagery, Sentinel-5P air quality data, and 14,000 social media weather report posts from Lagos to estimate daily heat island intensity maps. A Kalman filter fusion approach outperformed simple sensor averaging and kriging interpolation in predicting independently measured reference temperatures (RMSE 1.8 versus 2.7 degrees Celsius). Social media weather posts improved coverage in sensor-sparse areas of Lagos but introduced a systematic cool-hour reporting bias. Kano fusion experiments showed comparable accuracy with fewer available sensors. The study identifies three specific fusion methodology gaps for Nigerian urban climate applications and recommends NIMET partner with universities to develop a multi-modal urban climate monitoring network in Lagos and Kano as a research test bed.

Keywords: multi-modal data fusion, urban climate monitoring, Lagos, Kano, remote sensing