期刊
INTERNATIONAL JOURNAL OF WILDLAND FIRE
卷 22, 期 5, 页码 567-578出版社
CSIRO PUBLISHING
DOI: 10.1071/WF12177
关键词
Maxent; New Jersey; Ohio; Pennsylvania; predictive modelling; resource planning
类别
资金
- NorthSTAR
Wildfires are a serious threat for land managers and property owners, and over the last few decades this threat has expanded as a result of increased rural development. Most wildfires in the north-eastern US occur in the wildland-urban interface, those regions of intermingling urban and non-developed vegetated lands, where access to firefighting resources can be limited. We created monthly wildfire ignition probability maps from environmental predictors and wildfires occurring between 2000 and 2009 for the states of New Jersey, Ohio and Pennsylvania. Predictor variables included a drought index, long-term soil moisture, percentage forest and wildland-urban interface classifications. Probability maps generated from modelled (Maxent) extrapolations were used to create monthly hazard maps to aid agencies and managers with resource allocation and likelihood projections of wildfires across the region. Our results suggest that monthly hazard assessments provide a better indication of potential wildfires than does a single mean annual probability. Our monthly predictions retain information related to long-term seasonal variability associated with environmental variables and the recorded wildfires providing spatial and temporal information for resource allocation.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据