Spatio-temporal dynamics, drivers of wildfire occurrence and distribution in the northern savannah ecological zone of Ghana

Abstract

This study rigorously investigates the spatio-temporal dynamics and determinants of wildfire occurrences in the Northern Savannah Ecological Zone of Ghana from 2000 to 2021, leveraging remote sensing data and advanced statistical analyses. The study utilized Collection 6 MODIS datasets, including MCD64A1 for burned area mapping and MCD14DL for active fire locations. Temporal trends were analyzed using the Mann-Kendall test and Sen's slope estimator to detect significant changes. Over the 21 years, 432,153 active fires were recorded, resulting in a total burn area of 515,822.7 km². The peak wildfire occurrences were noted in 2011, with 28,943 fires, and in 2002, with a burn area of 33,883.4 km². Spatial analysis revealed concentrated wildfire hotspots in the northwest-central regions, while cold spots were primarily located in the Upper East region. Temporal trend analysis using the Mann-Kendall test indicated a significant decreasing trend in burn area over time (p = 0.015). Key drivers of wildfire occurrence and distribution were identified through Geographically Weighted Regression (GWR), which highlighted distance to settlements, slope, distance to roads, maximum temperature, and elevation as significant factors. The GWR model exhibited an improved fit over the global Ordinary Least Squares (OLS) model, as evidenced by a lower Akaike Information Criterion corrected (AICc) value, indicating enhanced model performance. The observed spatial heterogeneity in wildfire patterns underscores the necessity for localized modeling approaches and targeted management strategies. This study offers critical insights for the formulation of effective wildfire management policies in Ghana's Savannah zone, emphasizing the need to consider both environmental and anthropogenic factors in wildfire prevention and mitigation efforts.