Spatio-temporal clustering of active wildfire pixels over a 19-year period in a southern African savanna ecosystem of Zimbabwe

Abstract

ABSTRACT Understanding the spatial and temporal dynamics of active wildfires is critical for supporting policies and legislation related to wildfire management. The study tested the extent to which active wildfires significantly cluster in both space and time from 2000 to 2019 using space-time scan statistics permutation model (STSSP) available in the SaTScanTM software at the 6 month and annual scales. The statistical significance of each identified cluster was assessed using Monte Carlo simulation. In addition, the study assessed whether the radiative power of active wildfires exhibited significant trends over time using Mann-Kendall test. Results show that, over the 19-year period, primary clusters of active wildfires commonly occurred in the northern, southern and western districts of Zimbabwe. Significant clustering of active wildfires was observed between July and December at the 6-month scale and between January and December at the annual scale. Although mixed results were obtained regarding trends in radiative power of active wildfires, an increasing trend was observed across most districts and at the national scale suggesting that wildfires are getting hotter and covering large spatial extents. The ability to detect clustering of active wildfires in space and time provides rangeland managers with an opportunity for optimal and targeted management of wildfires.