Small-scale forest fire risk zoning based on bivariate statistics and multi-criteria decision analysis

Area of Science:

• Forestry
• Environmental Science
• Geographic Information Systems (GIS)

Background:

• Forest fires present significant risks to human life, ecosystems, and biodiversity.
• Effective forest fire management necessitates accurate, small-scale regional risk mapping.
• Previous methods may lack the comprehensive analytical power for precise risk assessment.

Purpose of the Study:

• To develop and evaluate novel comprehensive models for forest fire risk zoning.
• To integrate bivariate statistical methods with multi-criteria decision analysis for enhanced accuracy.
• To provide reliable decision support for forest fire management in Wangmo County, Guizhou Province.

Main Methods:

• Combined bivariate statistics (Weight of Evidence - WOE, Statistical Index - SI) with multi-criteria decision analysis (Analytic Hierarchy Process - AHP, Analytic Network Process - ANP).
• Developed and applied new WOE-ANP and SI-ANP comprehensive models for forest fire risk zoning.
• Validated model accuracy using existing data and compared the performance of ANP against AHP.

Main Results:

• Identified significant forest fire-prone areas in southern, western, and northern Wangmo County, with 39.2% of regions classified at level 4 risk or higher.
• The comprehensive WOE-ANP and SI-ANP models demonstrated high predictive accuracy (84.3% and 83.8%, respectively).
• The Analytic Network Process (ANP) provided more reliable weighting of forest fire risk factors compared to the Analytic Hierarchy Process (AHP).

Conclusions:

• The developed WOE-ANP and SI-ANP models significantly enhance the predictive capabilities for forest fire risk assessment.
• ANP offers a more robust approach for determining the relative importance of risk factors in forest fire analysis.
• These accurate and reliable models provide crucial decision support for effective forest fire management strategies.