Identifying forest thinning using anomalous change detection on synthetic aperture radar data

We apply anomalous change detection (ACD) to synthetic aperture radar (SAR) data to detect forest thinning at the Valles Caldera in New Mexico. By applying ACD across dimensions other than temporal, we establish baselines for change detection. Application of ACD across different polarizations highlights anomalous relationships associated with different types of scattering mechanisms. We also introduce a metric for distinguishing between anomalies consistently present in data over time and more subtle changes which may be obscured by these anomalies. This is especially useful for analyzing SAR backscatter intensity, which can be dominated by the presence of topographic features that are not of interest.