Remote Sensing and Machine Learning for Understanding Wildfire Regimes under Climate and Land Use Change: A Global Systematic Review with Implications for the South African Highveld

Global wildfire frequency and severity have increased due to the interaction between climate change and human-driven land use and land cover (LULC) changes, leading to widespread ecological, environmental, and socioeconomic impacts. Remote sensing has emerged as a crucial tool for investigating these changes, providing consistent, large-scale, and timely observations for wildfire monitoring and assessment. This systematic review analysed 128 peer-reviewed articles published between 2003 and 2023 to assess the impacts of climate change and LULC changes on wildfire occurrences and severity, with an emphasis on the South African Highveld. The bibliometric analysis revealed a substantial increase in research outputs over the study period. The United States contributed 49.2% (n = 63) of the reviewed articles, followed by China (10.2%) and Spain (9.4%). Landsat (43.1%), MODIS (21.3%), and Sentinel (16.8%) were the most frequently identified remote sensing platforms, while RF (28.7%) and SVM (19.5%) were the most frequently identified ML algorithms based on multi-label coding. The findings show that advances in Earth observation, cloud computing, and AI have improved wildfire detection, susceptibility mapping, burn severity, and predictive modelling. However, challenges such as data integration, geographical bias, transferability, and interpretability persist. In contrast, previous reviews examined climate change, LULC changes, remote sensing, or machine learning separately. This study offers a comprehensive synthesis of the interactions among climate change, LULC changes, remote sensing, and machine learning, while identifying emerging methodological gaps and highlighting future opportunities to develop more robust and operational wildfire monitoring. These findings offer valuable insights for researchers, land managers, and policymakers seeking to improve wildfire prediction, climate adaptation, and sustainable ecosystem management in a changing environment.