Dual-Polarimetric SAR Time-Series for Forest Disturbances Detection

Inhalt:
The future L-band NASA-ISRO SAR (NISAR) mission will provide global time-series of dual-polarimetric images every 6 days. This paper investigates the performance of a time-series approach for detecting forest disturbances caused by vegetation loss. Using a 3-year time-series of L-band ALOS-2 data collected over the Amazon rainforest, we show that vegetation removal has a distinct temporal pattern in the HV backscatter, which can be modeled by extending the Water Cloud Model with time-varying parameters. Fitting a generalized logistic function to the HV backscatter time-series leads to good disturbance detection accuracy when compared with the Hansen’s forest change map. This contribution confirms that SAR time-series analysis is an essential tool for monitoring forests and forest changes as time-series become increasingly available to the ecosystem and land-cover science community.