A Deep Learning Approach for Change Points Detection in InSAR Time Series

Abstract

Interferometric SAR (InSAR) algorithms exploit synthetic aperture radar (SAR) images to estimate ground displacements, which are updated at each new satellite acquisition, over wide areas. The analysis of the resulting time series finds its application, among others, in monitoring tasks regarding seismic faults, subsidence, landslides, and urban structures, for which an accurate and timely response is required. Typical analyses consist of identifying among the numerous time series the ones that exhibit an anomalous displacement, thus deserving to be further investigated. In practice, this is realised by selecting the time series which are characterised by trend changes w.r.t. the historical behaviour. In this work, we propose a Deep Learning approach for change point detection in InSAR time series. The designed architecture combines Long Short-Term Memory (LSTM) cells, to model the temporal correlation among samples in the input time series, and Time-Gated LSTM (TGLSTM) cells, to consider the sampling rate as additional information during learning. We further propose a solution to the lack of ground truth by developing a suitable pipeline for realistic data simulation. The method has been developed and validated through a large suite of experiments. Both quantitative and qualitative analyses have been conducted to demonstrate the detection capabilities of the learned model and how it is a valid alternative to the statistical reference algorithm. We further applied the developed method in a real continuous monitoring project to analyse InSAR time series over the Tuscany region in Italy, proving its effectiveness in the real domain.

Change Point Detection for Fine-Grained MFR Work Modes with Multi-Head Attention-Based Bi-LSTM Network

“…With its inspiration, this paper applied CNN-LSTM from anomaly detection in our MFR problem. The Bi-LSTM for radar pulse sequence analysis [ 31 ] and a version of Bi-LSTM involving weighted BCE loss function (Bi-LSTMw) [ 35 ] were both adopted as baselines.…”

Automatic Detection of Building Displacements Through Unsupervised Learning From InSAR Data

“…Chen et al [14] developed an LSTM neural network to predict land subsidence using time-series InSAR data, outperforming multilayer perceptron and recurrent neural network models, and enabling early warning and hazard relief. Lattari et al [13] also utilized LSTM cells and time-gated LSTM cells to monitor seismic faults, subsidence, landslides, and urban structure from nonuniformly sampled time-series.…”

“…An example of this is demonstrated by Anantrasirichai et al [17], who simulated the presence of slow deformation in both training and validation by creating a synthetic velocity map. Lattari et al [13] also employed a similar approach, augmenting their data with change points of varying slopes to simulate slow deformation. Additionally, Shakeel et al [15] added Gaussian peaks to the test data in order to simulate instant deformations caused by earthquakes.…”

“…On the other hand, the development of a machine learning (ML) or deep learning (DL) system for detecting anomalies in building displacements typically requires ground-truth data from in situ measurements [12] [13], such as displacement rate and type from reference buildings, which can be difficult to obtain due to the need for regular physical monitoring by experts. That is why the proposed solutions in this work rely on unsupervised representation learning based on the combination of autoencoders and long short-term memory (LSTM) networks to mitigate the lack of annotated data, similar to some recent efforts [14], [15].…”

Deep Learning-Based Landslide Recognition Incorporating Deformation Characteristics

“…Dai used SBAS InSAR to monitor the surface deformation area of Zhouqu County, analyzed its deformation characteristics and triggering reasons, and identified 23 active landslide deformation characteristics [26]. Lattari combined with long short-term memory to model InSAR time series and achieve continuous monitoring of landslides [27]. In addition, Cai proposed an integrated algorithm for landslide multi-source displacement optimization estimation based on the Kalman filter, which integrates displacement observation results from multiple platforms into a unified time series to achieve high temporal resolution monitoring of landslide motion [28].…”