Citation

Tanase MA, Kennedy R, Aponte C. Int. J. Wildland Fire 2015; 24(8): 1062-1075.

Copyright

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

DOI

10.1071/WF15059

PMID

unavailable

Abstract

Monitoring fire effects at landscape level is viable from remote sensing platforms providing repeatable and consistent measurements. Previous studies have estimated fire severity using optical and synthetic aperture radar (SAR) sensors, but to our knowledge, none have compared their effectiveness. Our study carried out such a comparison by using change detection indices computed from pre- and post-fire Landsat and L-band space-borne SAR datasets to estimate fire severity for seven fires located on three continents. Such indices were related to field-estimated fire severity through empirical models, and their estimation accuracy was compared. Empirical models based on the joint use of optical and radar indices were also evaluated. The results showed that optic-based indices provided more accurate fire severity estimates. On average, overall accuracy increased from 61% (SAR) to 76% (optical) for high-biomass forests. For low-biomass forests (i.e. aboveground biomass levels below the L-band saturation point), radar indices provided comparable results; overall accuracy was only slightly lower when compared with optical indices (69% vs 73%). The joint use of optical and radar indices decreased the estimation error and reduced misclassification of unburned forest by 9% for eucalypt and 3% for coniferous forests.

Language: en