期刊
REMOTE SENSING OF ENVIRONMENT
卷 107, 期 4, 页码 533-544出版社
ELSEVIER SCIENCE INC
DOI: 10.1016/j.rse.2006.10.001
关键词
fire risk; static models; vegetation characterization; logistic regression; fire history
A detailed understanding of the spatial patterns of burning is valuable for managing biodiversity and ecosystems. This research assesses the performance of several spectral indices derived from Landsat data when modelling fire occurrence probability by means of logistic regression. Normalized Difference Vegetation Index (NDVI), Normalized Difference Moisture Index (NDMI), Normalized Burn Ratio (NBR) and the greenness and wetness components of the Tasseled Cap Transformation were tested. Landscape variables (topography, accessibility and structural vegetation) were also included as predictors in models development. Although fire fisk is closely related to weather and vegetation status at a given time, it is also strongly linked to fire history, and changes in predictor values in years previous to the fire were events also considered. The models generated correctly classified about 70% of the validation data set. The inclusion of pre-fire spectral indices improved models ability to predict fire occurrence. Although the NBR-based model was the most accurate, TCWetness and NDVI-based models showed similar results, while TCGreenness performed worst. Models with no spectral indices described the fire-proneness of the landscape structure, while the inclusion of spectral indices improved the recognition of particular spatial conditions. Slope and distance to the nearest path were also identified as valuable predictors. All the models identified the main fire risk zones in the study area. Their integration into a single, integrated model properly described fire-proneness and is suggested to be a valuable tool for the identification and management of fire risk. The method used is simple, describes the key variables and spatial pattern of the fire regime and is suited to operational use in Mediterranean ecosystems. (c) 2006 Elsevier Inc. All rights reserved.
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