Spatiotemporal evolution characteristics and driving mechanisms of wildfires in China under the context of climate change and human activities

Abstract

With the ongoing global warming and frequent droughts, the threat of wildfires is also increasing. For instance, the recent wildfires in Los Angeles, California, have not only destroyed homes, including those of several billionaires and celebrities, but also resulted in multiple fatalities, drawing widespread international attention. Similarly, China, located at the same latitude as the United States but on the other side of the globe, is also facing an escalating wildfire threat. However, the current research on wildfires in China remains insufficient, particularly regarding their spatiotemporal evolution and driving factors. Therefore, this study, based on MODIS satellite data and interpretable machine learning model, systematically analyzed the spatiotemporal variations and driving forces of wildfires in China from 2001 to 2020, revealing the impacts of climate change and human activities on wildfire occurrence. The results show that: (1). Over the past two decades, the overall frequency of wildfires in China has shown an increasing trend followed by a decrease. Spatially, the frequency and intensity of wildfires were significantly higher in the eastern regions compared to the western inland areas. Temporally, wildfires were concentrated mainly in the spring season and during the daytime. Additionally, there was a notable time-lag and cumulative effect in wildfire occurrences. (2). The frequency and intensity of wildfires in China generally exhibited a high degree of consistency (R2: 0.846), although there were some discrepancies in specific regions. (3). Climate factors were the dominant drivers of wildfire occurrence in China (48 %), with temperature having a particularly significant effect. Notably, the risk of wildfires increased markedly when temperatures exceed 16 °C. (4). In addition to climate factors, human activities also played a crucial role in the occurrence of wildfires, particularly cultural traditions and festivals in China (e.g., Qingming Festival), which may significantly influence wildfire incidence. This study systematically revealed the spatiotemporal differentiation mechanism of wildfires in China under the dual influence of climate change and human activities, along with their nonlinear response patterns. The study suggests that China should prioritize the establishment of seasonal and region-specific wildfire early warning systems and strengthen fire prevention measures during high-risk periods such as spring and traditional festivals.

Keywords

• China wildfires
• Spatial-temporal characteristics
• Driving factors
• Interpretable Machine Learnin
• eXtreme Gradient Boosting (XGBoost)
• Shapley Additive Explanations-Partial Dependence Plots (SHAP-PDP)