Land cover change effects on land surface temperature trends in an African urbanizing dryland region

Abstract

Land use-land cover (LULC) association with land surface temperature (LST) is well known. Knowledge about land change effects on LST in urbanizing African drylands is very limited. We examined LST and vegetation variations in semi-arid Gaborone (Botswana's capital) using MODIS daytime and night-time LST (DNLST), and Normalized Difference Vegetation Index (NDVI) between 2000 and 2018. Significant land transitions were identified in the land cover change map using Change Vector Analysis of Landsat-based biophysical indices of vegetation, water and bare soil. Artificial surface and tree-covered areas were net gaining categories, whereas cropland and grassland were net losing categories. Detailed profiling of DNLST trends and breakpoints was conducted in five relatively homogenous sites representing land cover/transitions. Increasing NDVI and DNLST trends found were significant. Per class, LST change at daytime and night-time are as follows: built-up areas (1.8 K, 2.2 K), Gaborone dam (5.7 K, 0.2 K), settlement expansion areas (4.6 K, 2.2 K), and rural settlement (2.0 K, 1.5 K). The cooling effect of irrigation on daytime LST was higher than night-time LST as daytime LST trend as low as −0.4 K was found in areas of irrigated croplands. Validation with synoptic station temperature data and dam water levels provides empirical evidence that MODIS gave credible DNLST estimates in this urbanizing dryland area. Our results also suggest the role of climate variability in urbanizing drylands alongside land cover change in controlling the LST. Regardless, coupling DNLST and land cover changes can provide useful information for spatial planning of drylands to create smart cities that are resilient to climate change.