Spatial model of wildfire susceptibility using Machine Learning approaches on Rawa Aopa Watumohai National Park, Indonesia

Author:

Aldiansyah Septianto1,Madani Ilyas2

Affiliation:

1. Department of Geography , Faculty of Mathematics and Natural Sciences, Universitas Indonesia , Depok , Indonesia

2. Department of Geography , Faculty of Social Sciences, Universitas Negeri Malang , Malang , Indonesia

Abstract

Abstract Rawa Aopa National Park has experienced a severe wildfire. These fires are affected by several factors, including topography, meteorology, vegetation, and source of fire. This study uses a Machine Learning approach based on re-sampling methods (e.g. crossvalidation, bootstrap, and random subsampling) to evaluate, and improve the performance of twelve basic Machine Learning algorithms: Generalized Linear Model, Support Vector Machine, Random Forest, Boosted Regression Trees, Classification And Regression Tree, Multivariate Adaptive Regression Splines, Mixture Discriminate Analysis, Flexible Discriminant Analysis, Maximum Entropy, Maximum Likelihood, Radial Basis Function, and Multi-Layer Perceptron, analyze the causes of wildfires, and the correlation between variables. The model is evaluated by Area Under Curve, Correlation, True Skill Statistics, and Deviance. The evaluation results show that Bt-RF has a good performance in predicting wildfire susceptibility in TNRAW with AUC=0.98, COR=0.96, TSS=0.97, and Deviance=0.15. An area of 644.88 km2 or the equivalent of 59.82% of the area is a wildfire susceptibility area with the concentration of fires occurring in the savanna ecosystem which is around 245.12 km2 or the equivalent of 88.95% of the jungle zone. Among the 17 parameters that cause fires, this area is strongly influenced by Maximum Temperature, Land Use Land Cover, and Distance from Road. There is a strong correlation between soil and distance from settlements = 0.96.

Publisher

Walter de Gruyter GmbH

同舟云学术

1.学者识别学者识别

2.学术分析学术分析

3.人才评估人才评估

"同舟云学术"是以全球学者为主线,采集、加工和组织学术论文而形成的新型学术文献查询和分析系统,可以对全球学者进行文献检索和人才价值评估。用户可以通过关注某些学科领域的顶尖人物而持续追踪该领域的学科进展和研究前沿。经过近期的数据扩容,当前同舟云学术共收录了国内外主流学术期刊6万余种,收集的期刊论文及会议论文总量共计约1.5亿篇,并以每天添加12000余篇中外论文的速度递增。我们也可以为用户提供个性化、定制化的学者数据。欢迎来电咨询!咨询电话:010-8811{复制后删除}0370

www.globalauthorid.com

TOP

Copyright © 2019-2024 北京同舟云网络信息技术有限公司
京公网安备11010802033243号  京ICP备18003416号-3