from diffuse subsidence to persistent net subsidence by 2023, with the strongest and most heterogeneous settlement in the West sector and along the Crest during active construction.. LightGBM provided useful short-horizon forecasts of one-step vertical deformation increments, with low zone-wise errors relative to cumulative settlement magnitude. Performance was strongest in the more stable sectors and weaker where deformation was more heterogeneous. The main limitations were transient anomalous acquisitions, construction-related non-stationarity, and the absence of operational covariates. Nevertheless, the results support MT-DInSAR-based forecasting as a complementary tool for localised TSF monitoring, particularly where predicted increments are used to detect changes in deformation velocity and identify localised acceleration. Its utility as an early-warning tool could be maximised by integrating model outputs into existing operational monitoring systems, where satellite-based forecasts are routinely cross-checked against ground-based observations to support earlier and better-informed preventive site management. ACKNOWLEDGEMENTS The researchers would like to acknowledge the Science and Technology Research Partnership for Sustainable Development (SATREPS), Japan Science and Technology Agency (JST)/Japan International Cooperation Agency (JICA) for the support to the Japan-Kazakhstan SATREPS Knight Project during this research. REFERENCES Bansal, Ms. A., Sharma, Dr. R., & Kathuria, Dr. M. (2022). A Systematic Review on Data Scarcity Problem in Deep Learning: Solution and Applications. ACM Computing Surveys, 54(10s), 1– 29. https://doi.org/10.1145/3502287 Canadian Dam Association. (2013). Dam Safety Guidelines. Chen, J., Sun, J., Xia, Y., Xiong, F., Li, X., Liu, C., Hu, Y., & Shao, C. (2025). Intelligent Prediction Based on NRBO–LightGBM Model of Reservoir Slope Deformation and Interpretability Analysis. Water, 17(22), 3248. https://doi.org/10.3390/w17223248 Crosetto, M., Monserrat, O., Cuevas-González, M., Devanthéry, N., & Crippa, B. (2016). Persistent Scatterer Interferometry: A review. ISPRS Journal of Photogrammetry and Remote Sensing, 115, 78–89. https://doi.org/10.1016/j.isprsjprs.2015.10.011 Das, S., Priyadarshana, A., & Grebby, S. (2024). Monitoring the risk of a tailings dam collapse through spectral analysis of satellite InSAR time-series data. Stochastic Environmental Research and Risk Assessment, 38(8), 2911–2926. https://doi.org/10.1007/s00477-02402713-3 F. Gama, F., Mura, J. C., R. Paradella, W., & G. de Oliveira, C. (2020). Deformations Prior to the Brumadinho Dam Collapse Revealed by Sentinel-1 InSAR Data Using SBAS and PSI Techniques. Remote Sensing, 12(21), 3664. https://doi.org/10.3390/rs12213664 Ge, Q., Wang, J., Liu, C., Wang, X., Deng, Y., & Li, J. (2024). Integrating Feature Selection with Machine Learning for Accurate Reservoir Landslide Displacement Prediction. Water, 16(15), 2152. https://doi.org/10.3390/w16152152 Hanssen, R. F. (2001). Radar Interferometry: Data Interpretation and Error Analysis (Vol. 2). Springer Science & Business Media.
RkJQdWJsaXNoZXIy MTM0Mzk2