Urban Vegetation Cooling Efficiency Along the Urban–Peri-Urban Gradient: A Multi-Temporal Remote Sensing Assessment of Gujranwala, Pakistan

Abstract

Urban vegetation is widely reported to lower land surface temperature (LST), but whether the magnitude of this cooling effect varies with position along the urban–peri-urban gradient, and whether the spatial configuration of vegetation patches matters independently of vegetation amount, remains poorly documented for Pakistani cities. This study examined Gujranwala, Punjab, using Landsat 5, 8, and 9 Collection 2 Level-2 imagery processed in Google Earth Engine across four epochs (2000, 2010, 2020, and 2025). An empirically delineated urban core, based on Dynamic World built-up probability, anchored four concentric distance zones extending to 15 kilometers. LST, NDVI, NDBI, and MNDWI were computed for each epoch and aggregated to a 250-meter grid to limit the inflated statistical significance associated with spatial autocorrelation. NDVI was strongly and negatively associated with LST (coefficient − 16.54, p < 0.001, R² = 0.835), and NDBI was strongly positively associated with LST (coefficient 27.50, p < 0.001, R² = 0.871), consistent with the wider literature. Cooling intensity, defined relative to a local built-up reference temperature, was higher in the urban core and inner-urban zones than in the outer fringe and peri-urban zones in three of the four epochs. Vegetation patch configuration, tested through the Largest Patch Index and patch density, added only a small amount of explanatory power beyond vegetation amount alone (R² increase of approximately 0.06 percentage points), with patch contiguity associated with cooling in the expected direction. A systematic, epoch-level offset in mean LST was identified and controlled for using year fixed effects in all regression models. These findings indicate that vegetation amount remains the dominant determinant of surface cooling in Gujranwala, while its spatial arrangement contributes a modest and only partially consistent additional effect.