Interferometric SAR Coherence Variations With Amazon Floodplain and Lake Vegetation

Interferometric processing of synthetic aperture radar (SAR) data has been demonstrated to yield cm-scale measurements of water level changes within flooded vegetation. The method relies upon vegetation to provide a double-bounce travel path, yet descriptions of the vegetation types and related coherence are not prevalent. We demonstrate that interferometric processing of JERS-1 SAR data over Lake Balbina within the Amazon Basin shows coherence changes in eight identifiable ecological classes: open water, bare soil/non-flooded herbaceous, aquatic macrophytes, non-flooded shrub, flooded shrub, flooded woodland, non-flooded forest, and flooded forest. The classes are applicable for both low and high water images. Using an interferogram with a short temporal baseline of 44 days (perpendicular baseline of 293 meters), coherence of the flooded woodland and non-flooded forest are 0.49 and 0.30, respectively (for the low water levels). For high water levels, coherence values of 0.46 and 0.31, respectively, were found. Alternatively, when using an interferogram with a long temporal baseline of 616 days (perpendicular baseline of 1238 meters), coherence of the flooded woodland and non-flooded forest are 0.21 and 0.15 at low water levels. Our preliminary results suggest that as temporal and spatial baselines increase, coherence values decrease. We will continue to study interferograms of various temporal and spatial baselines for their coherence values over Amazonian lakes and wetlands.