Modeling Land Surface Temperature Dynamics Using Biophysical and Environmental Drivers for Prescribed Fire Planning in Serengeti National Park

Abstract

Despite the extensive research on the factors that influence Land Surface Temperature (LST), the combined effects of vegetation structure, soil moisture, land cover dynamics, and fire activity on surface temperature regulation remain underquantified across heterogeneous protected landscapes such as national parks. This study investigated the relationship between LST and multiple biophysical and environmental variables, including the Normalized Difference Vegetation Index (NDVI), burn severity (dNBR), soil moisture, and land cover, in Serengeti National Park, Tanzania, from 2001 to 2024, to support datadriven fire and habitat management strategies. Model evaluation metrics (R², MAE, RMSE, MAPE, BAE, and MBE) indicated high predictive performance, and residual analysis confirmed model validity and homoscedasticity. Grasslands overwhelmingly dominated the landscape (≈99%), while savannas expanded slightly and croplands declined after 2006. Burn severity analysis based on the dNBR classification revealed predominantly low to moderate severity, with limited areas of high severity concentrated in the western and central zones. Areas of higher burn severity corresponded to elevated LST, while unburned or lightly burned regions exhibited cooler temperatures and higher vegetation recovery potential. Soil moisture showed pronounced seasonal and interannual variability, with surface layers responding rapidly to rainfall, and deeper layers displaying greater stability and moisture retention. Peaks in 2006–2007, 2017–2018, and 2020–2021 coincided with wetter conditions, whereas declines occurred during prolonged dry periods. The results demonstrated a strong inverse relationship between NDVI and LST, highlighting vegetation’s cooling effect and its role in post-fire recovery. The findings emphasize the importance of integrating LST and its biophysical predictors into fire and ecosystem management frameworks to enhance ecological resilience and guide adaptive, climate-informed decision-making in protected savanna landscapes.