ABSTRACT
In this study, we evaluated the usefulness of Sentinel-1A&B Synthetic Aperture Radar (SAR) data for burnt area mapping at a country scale in Laos. We used both the VV) and VH) polarization data to assess the burnt areas and other land cover signal variations. We also compared the burnt area backscatter signal variations with incident angles. Results suggested VV having a relatively higher backscatter than VH for all land cover classes and months (March, April, and May, 2018 and 2019). Water and barren areas had a distinct VV and VH backscatter signal compared to the other land cover classes. We observed a consistent pattern in the VV and VH backscatter signal with the highest to lowest backscatter in the following order: forests>burnt areas>urban areas>croplands>barren>water, during March with other slight variations in other months. The VH signal difference was distinct between burnt areas versus croplands and burnt areas versus urban areas in March and April, which are the peak biomass burning months. As the incident angle increased beyond 40°, the data were highly scattered for VV and VH burnt areas; however, both signals were unique. Mapping burnt areas using VH yielded relatively higher accuracy than VV polarization. Our results highlight the burnt area signal variations from Sentinel-1A&B SAR data in Laos useful for mapping and monitoring purposes.
