Analysis of seasonal and long-term anomalies in vegetation from coregistered SAR stacks

Abstract

This thesis analyses anomalous seasonal and long-term effects in time series results of applying Incoherent Cross-Correlation (ICC) on co-registered stacks of Synthetic Aperture Radar (SAR) images, demonstrating possible correlation to forest growth and in-plant water dynamics. It is motivated by the fact that forest parameters like height and structure are important in many applications related to biomass calculation, which is crucial for finding climate change mitigation solutions. However, there are many challenges in estimating these parameters since some forests are not readily accessible and are in large areas, while many traditional ground-based methods lack scalability to provide persistent largescale monitoring. With the advancement of space-borne platforms, this work analyses results from the well-established ICC technique novelly applied to forest growth monitoring using a multi-year SAR stack to achieve centimetre-level range shifts. This work will also attempt to provide possible explanations as to why range shifts are mainly observed in the forest backscatter but not in Persistent/Distributed Scatterer (PS/DS), and how these range shifts correlate with both sensor and terrain specific factors. To our current knowledge, ICC applied on large scale forest monitoring using active sensors like Sentinel-1 represents a novel discovery. Radar delay could now be used to track forest growth and inter-annual water content variability that can provide insights into vegetation health and drought stress, which were previously impossible using conventional PS/DS approaches that rely solely on phase-based coherence.