We present the preliminary results of an InSAR analysis of peatland surface motion covering a large spatial and temporal extent. This work is the first large scale analysis of the Dutch Green Heart region, and is made possible using a novel distributed scatter (DS) InSAR processing method. This method is designed to handle breakages in the observed interferometric phase time series which occur due to temporal decorrelation, which we designate with the term loss-of-lock.

We present a methodology which enables InSAR observations of ground motion in regions of low coherence and periodic decorrelation events which makes use of spatial and temporal contextual data to increase the information available to the processing algorithms. While this study is focused on observations of cultivated peatlands, the approach is generic and can be applied to other types of regions where coherence loss may be a concern. Neighbouring regions are grouped together by their contextual attributes such that when one region decorrelates, other observations from similarly behaving regions can still be used to derive a kinematic displacement model, thereby spanning the incoherent gap.

We introduce the termloss-of-lockto describe a specific form of coherence loss which results in the breakage of an InSAR time series. Loss-of-lock creates a specific pattern in the coherence matrix of a multilooked distributed scatterer (DS) by which it may be detected. Along with identification, we introduce a new DS processing methodology which is designed to mitigate the effects of loss-of-lock by introducing contextual data to assist in the time series processing. This methodology is of particular relevance to regions which suffer from severe temporal decorrelation, such as northern peatlands.We apply our new method to two subsiding cultivated peatland regions in The Netherlands which previously proved impossible to monitor using DS InSAR techniques. Our results show a very good agreement with in-situ validation data as well as spatial correlation between regions and the natural terrain.

We present a novel InSAR processing scheme which combines point scatterer (PS) and distributed scatter (DS) approaches in a hybrid framework along with contextual information about the environment under study. Data such as land parcel divisions, precipitation and temperature are integrated into the processing pipeline in order to produce accurate deformation time series estimates of the Dutch peatlands. In addition to these steps, a segmented processing scheme is introduced to manage irreversible losses of coherence in the interferogram stack. Initial results show a promising agreement with in-situ ground truth measurements gathered by extensometer readings of shallow surface deformation.

Phase unwrapping, also known as ambiguity resolution, is an underdetermined problem in which assumptions must be made to obtain a result in SAR interferometry (InSAR) time series analysis. This problem is particularly acute for distributed scatterer InSAR, in which noise levels can be so large that they are comparable in magnitude to the signal of investigation. In addition, deformation rates can be highly nonlinear and orders of magnitude larger than neighboring point scatterers, which may be part of a more stable object. The combination of these factors has often proven too challenging for the conventional InSAR processing methods to successfully monitor these regions. We present a methodology which allows for additional environmental information to be integrated into the phase unwrapping procedure, thereby alleviating the problems described above. We show how problematic epochs that cause errors in the temporal phase unwrapping process can be anticipated by the machine learning algorithms which can create categorical predictions about the relative ambiguity level based on the readily available meteorological data. These predictions significantly assist in the interpretation of large changes in the wrapped interferometric phase and enable the monitoring of environments not previously possible using standard minimum gradient phase unwrapping techniques.

The Dutch peatlands are a notoriously difficult region to monitor using InSAR. Low temporal coherence and signal-to-clutter levels necessitate the extraction of collective behaviour by the suppression of noise and clutter. Conventional techniques used to accomplish this include multilooking and phase-linking. The t-distributed Stochastic Neighbour Embedding (t-SNE) algorithm is a dimensionality reduction technique that aids in the analysis of large datasets. In this paper, we present an initial investigation into the suitability of the t-SNE algorithm to take the idea of extracting collective behaviour further. Similarly-behaved patches of land are automatically grouped together by the algorithm which aids in the specification of a functional model for that group. Our initial results show that the algorithm is able to successfully identify and group together areas in a scene which display similar behaviour over time. We also find that groups which display the same behaviour may also contain the same kinds of processing errors (for example unwrapping errors or cycle slips) and that these can also be automatically detected by the algorithm. We present this result as the first building block in an approach to smart InSAR data analysis which can learn from the data it is processing.