SPATIAL REGRESSION OF AIR QUALITY AND LAND SURFACE TEMPERATURE IN CAGAYAN DE ORO AIRSHEDS

This study analyzes the spatial relationship between Land Surface Temperature (LST), Elevation, Normalized Difference Vegetation Index (NDVI), Normalized Difference Built-up Index (NDBI), and key air pollutants (NO₂, SO₂, CO) across the Metro Cagayan de Oro Airsheds, encompassing Cagayan de Oro City and the surrounding municipalities of Opol, El Salvador, Tagoloan, Villanueva, and Jasaan, Philippines. Satellite-derived MODIS LST and Sentinel-5P TROPOMI air quality datasets were spatially harmonized and resampled to a uniform 1km spatial grid. Statistical screening via Pearson correlation (r) revealed weak positive global relationships for NO₂ (r = +0.2137) and NDBI (r = +0.2137), while CO, NDVI, and SO_2, and elevation exhibited negligible global associations. Diagnostic tests for Ordinary Least Squares (OLS) regression demonstrated significant global limitations (R² = 0.3169, p > 0.0001), with critical failures including spatial autocorrelation (Durbin-Watson = 1.075), non-normality (Jarque-Bera p < 0.0001), heteroskedasticity (Koenker/BP p < 0.0001), and severe multicollinearity (Condition Number = 6.80 x 10⁷). To account for spatial non-stationarity across the airshed, Geographically Weighted Regression (GWR) was applied, substantially improving explanatory performance with mean local R² = 0.433 and a peak local R² = 0.670. GWR local coefficients revealed pronounced spatial volatility and directional reversals—most notably for NDBI (ranging from -134.10 to +523.41) and DEM (-0.0493 to +0.0373). These empirical findings confirm that global spatial models misrepresent microclimatic drivers, underscoring the necessity of localized, GWR-derived spatial analytics for evidence-based urban heat management and airshed policy across Metro Cagayan de Oro.