Mapping Grassland Wildfire Risk of the World

Abstract

Wildfire (forest fire and grassland fire) is a natural process that disturbs the ecosystem succession and has important impacts on regional ecosystem, global climate cycle, and local society and economy. The grassland fire risk assessment model involves the grassland burning probability and the potential economic loss. A multivariate logistic regression model was used for modeling grassland burning probability, using remotely sensed vegetation indices as the proxies of fuel properties and the topographic factors as the explanatory variables, and MODIS burned areas as the response variable. The potential economic loss uses NPP (Net Primary Product) as a surrogate to represent the potential loss of grassland fire. Based on the grassland fire risk assessment model, we first calculated the grassland burning probability at 8-day scale for 2000{\textasciitilde}2010, and then calculated the annually averaged grassland burning probability. The yearly grassland fire `risk' was then modeled by the product of yearly grassland burning probability and NPP. The final global grassland burning probability map and risk map was obtained by averaging the above results during 2000{\textasciitilde}2010 with 1 km spatial resolution. The grassland fire risk mapping result indicates that the high grassland fire risk regions are distributed in the central part of Asia, western Europe, southwestern Africa, northern Oceania, central part of North America, and northeastern South America.