Analysis of Spatial-Temporal Patterns of Wildfire Susceptibility in Queen Elizabeth National Park (QENP) - Uganda

Abstract

This study determined the variability of wildfire susceptibility in Queen Elizabeth National Park (QENP) in space and time. QENP is a protected area in Western Uganda. MODIS and VIIRS data for a six-and-a-half-year period from January 2015 - June 2021 were obtained to create an inventory of past fires. From these fires, spatial and temporal patterns were derived from exploratory spatial data analyses. The Weights of Evidence (WOE) method, a Bayesian form of statistical modelling, was used to determine the relationship between fires and wildfire conditioning factors, as well as to model wildfire susceptibility. Results of the study showed that the occurrence of wildfires within the study area vary seasonally. Sixty-one percent (61%) of the fires were observed to occur in the first dry season of the year, while thirty-one percent (31%) of the fires were observed to occur in the second dry season. Among the wildfire conditioning factors, altitude, vegetation (as measured by NDVI), and proximity to lakes indicated the highest correlation with the occurrence of fires. These conditions were attributed to physiographic influences, water stress in vegetation, and the socio-economic activities of the fishing villages around the lakes respectively. From the derived wildfire susceptibility maps, varying levels of wildfire susceptibility were determined. Proportional values of 19% and 20% of the study area were classified with very high and high susceptibility levels respectively. The remaining 61% of the study area was covered by moderate, low, and very low susceptibility levels. The study results provided vital findings about the seasonal patterns of wildfire occurrence, factors influencing the occurrence of wildfires and the locations most susceptible to wildfires. This information will enable managers to allocate fire management resources optimally to efficiently mitigate against wildfires within QENP.