Wildfire Risk Assessment and Zoning by Integrating Maxent and GIS in Hunan Province, China

• Published in: Forests 2021-10, Vol.12 (10), p.1299
• Main Authors: Yang, Xuhong, Jin, Xiaobin, Zhou, Yinkang
• Format: Article
• Language: English
• Subjects: Aridity; Climate change; Climatic conditions; Elevation; environment factors; Environmental conditions; Environmental effects; Environmental risk; Fire prevention; Forest & brush fires; Forest ecosystems; Geographic information systems; Geographical variations; High temperature; International organizations; Land use; Machine learning; Maxent model; Neural networks; Observatories; Population density; Radiation; Remote sensing; Risk assessment; Sensors; Terrestrial ecosystems; Timber; Topography; Variables; Vegetation; Vegetation cover; wildfire; Wildfires; Wind speed; Zoning
• ISSN: 1999-4907; 1999-4907
• DOI: 10.3390/f12101299

Forest wildfire is an important threat and disturbance facing natural forest ecosystems. Conducting wildfire risk assessments and zoning studies are of great practical significance in guiding wildfire prevention, curbing fire occurrence, and mitigating the environmental effects of wildfire. Taking Hunan Province, China as the case area, this study used remotely sensed high-temperature fire data as the wildfire sample. Twelve factors related to topography, climatic conditions, vegetation attributes, and human activities were used as environmental variables affecting wildfire occurrence. Then, a Maxent wildfire risk assessment model was constructed with GIS, which analyzed the contribution, importance, and response of environmental variables to wildfire in Hunan Province. The results show that (1) the Maxent model has high applicability and feasibility when applied to wildfire risk assessment after a test of wildfire sample sites; (2) the importance of meteorological conditions and vegetation status variables to wildfire are 54.64% and 25.40%, respectively, and their contribution to wildfire are 43.03% and 34.69%, respectively. The interaction between factors can enhance or weaken the contribution of factors on wildfire. (3) The mechanism for the effects of environmental variables on wildfire is not linear as generally believed; temperature, aridity, land use type, GDP, distance from the road, and population density have a nonlinear positive correlation with the probability of wildfire occurrence. Elevation, slope, precipitation, wind speed, and vegetation cover within the suitable interval positively contribute to the probability of wildfire, while the environmental conditions outside the suitable interval curb the occurrence of wildfire. The response of wildfire probability to forest density is U-shaped, which means either too high or too low will promote the occurrence of wildfire. (4) There is geographical variation of wildfire risk in Hunan Province. The areas at high risk and below account for 74.48% of the total area, while the areas at significantly high risk and above account for a relatively low proportion, 25.52%.