Wildfire Progression Time Series Mapping With Interferometric Synthetic Aperture Radar (InSAR)

Abstract

We describe a novel algorithm to accurately characterize burned area and generate a time series of active burned areal extent during an actively burning wildfire based upon changes in the second-order statistics of interferometric synthetic aperture radar (InSAR) phase measurements. We present this algorithm and demonstrate its use with Sentinel-1 InSAR data collected during the 2020 Cameron Peak Fire, which burned along the Front Range in Colorado, USA. We show that this algorithm can successfully discriminate recently burned and actively burning areas within a fire zone from unburned areas at high spatial resolution (~10s of m). We further introduce a method for estimating a time series of burned areal extent from interferometric observations of burned area-change via a singular value decomposition (SVD) inversion. We compare the results of our algorithm with fire progression maps from the National Interagency Fire Center (NIFC) and find good agreement on total burned area (IoU=0.65) and excellent agreement on burned area extent (mIoU=0.91).