شناسايي نقطه تغيير در سري‌هاي زماني حاصل از تداخل‌سنجي راداري با استفاده از شبكه عصبي عميق، منطقه مطالعاتي: قاره اروپا

The Earth’s surface is continuously deformed by natural an‎d anthropogenic processes, an‎d such deformations can potentially evolve into severe geohazards. Interferometric Synthetic Aperture Radar (InSAR) has emerged as a key geodetic technique for precise an‎d frequent monitoring of ground displacements over large spatial scales, particularly with the advent of spaceborne constellations such as Sentinel-1. Nevertheless, the automatic an‎d robust detection of change points (CPs) in InSAR displacement time series remains highly challenging due to the presence of severe atmospheric noise, non-linear deformation patterns, data gaps, an‎d irregular sampling intervals. These limitations substantially constrain the performance of conventional statistical an‎d model-based approaches. The primary objective of this dissertation is to develop a comprehensive deep learning–based framework for automatic change point detection (CPD) in InSAR time series. To this end, a novel synthetic data generation pipeline was designed that realistically models’ displacement trends (linear an‎d Mogi-type), seasonal components (semi-annual an‎d annual sinusoidal signals), an‎d multiple types of noise (Gaussian, Poisson, an‎d white). To ensure realism an‎d avoid manual intervention, an automated filtering procedure was implemented using a Multi-Layer Perceptron (MLP), Dynamic Time Warping (DTW), an‎d DBSCAN clustering. Building upon this foundation, four advanced deep neural architectures were developed: MLP, providing computational simplicity an‎d efficiency; MALkCNN, a large-kernel convolutional neural network optimized for extracting long-term trends an‎d reducing local fluctuations; ATGLSTM, an Attention–Time-Gated LSTM designed to maintain robustness under noisy an‎d gap-filled conditions; an‎d CNN–LSTM, a hybrid architecture that leverages convolutional layers for spatial feature extraction an‎d LSTM units for temporal dependency modeling. The proposed models were validated on both synthetic datasets an‎d real-world InSAR time series acquired over geodynamically active regions in Europe, including Germany, Italy (Campi Flegrei caldera), an‎d Icelan‎d. Quantitative eva‎luations demonstrated outstan‎ding performance: MLP achieved ~97% accuracy while reducing computational cost by nearly an order of magnitude compared to deeper models. MALkCNN reached an F1-score of ≈0.98 on synthetic data an‎d ≈83% on real data, operating ~7× faster than the TG-LSTM baseline. ATGLSTM maintained high stability in gap-affected an‎d noisy datasets, yielding F1 ≈72% on Icelan‎dic time series. CNN–LSTM attained ‎>95% accuracy on synthetic benchmarks an‎d strong correlations with GNSS records at Campi Flegrei (Pearson’s ρ = 0.88 for ascending an‎d ρ = 0.85 for descending tracks). These results confirm that the proposed deep learning framework substantially outperforms classical approaches in terms of detection accuracy, robustness to data irregularities, an‎d computational scalability. Moreover, the models produced spatio-temporal change maps consistent with known geophysical processes such as magma migration, hydrothermal pressurization, an‎d isostatic adjustment, thereby bridging computational advances with geophysical interpretation. Overall, this dissertation demonstrates that advanced deep learning architectures, when trained on realistic synthetic datasets an‎d validated against independent geodetic observations, constitute a powerful an‎d scalable solution for automatic, reliable, an‎d near-real-time CPD in InSAR time series. The framework not only advances methodological research in radar remote sensing but also provides a practical foundation for the development of operational early-warning systems for volcanic unrest, subsidence, an‎d other geohazards.

7

150630