Citation

Chu T, Guo X. Int. J. Wildland Fire 2015; 24(3): 419-432.

Copyright

(Copyright © 2015, International Association of Wildland Fire, Fire Research Institute, Publisher CSIRO Publishing)

DOI

10.1071/WF14124

PMID

unavailable

Abstract

Wildfire is the main natural disturbance in forest ecosystems; it controls and modifies vegetation compositions, landscape properties and global carbon cycle. Estimates of areas burned by wildfires vary greatly depending on the environmental conditions, data availability and methods used. This paper aims to develop a framework for reconstructing time series of burned areas in the taiga-steppe transition zone using MODIS composites. The estimated accuracy of the developed mapping algorithm and other statistical indications denote that the clear land surface composites of MODIS data in spring (Julian dates, JD 97-177), logistic regression and MODIS active fire product can be integrated successfully for reconstructing burned areas in the taiga-steppe transition zone. Time series of burned areas between 2000 and 2012 derived from the MODIS spring composite algorithm were validated using Landsat-based burned areas, showing average omission and commission errors of 18% and 31%. Compared with the MCD45A1 burned area product, the developed algorithm significantly improved the prediction of burned areas and successfully separated late-season from early-season burns. The derived long-term burned areas will assist in understanding the complex relationships among forest dynamics, forest recovery and fire in the vulnerable boreal forest ecosystem as well as its transition zone under climate change in northern Mongolia and Central Asia.

Language: en