EUVIMEDEUVIMED entdecken
Crossref

Hierarchical system for forecasting and decomposition analysis of hydraulic structure deformations based on geodetic time series

Oleksandr Hrabovyi

Geodesy, cartography and aerial photography · 2026-05 · 103,2026 (103) · 66-80

DOI: 10.23939/istcgcap2026.103.066 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Geodesy, cartography and aerial photography, 2026-05. Verfasst von Oleksandr Hrabovyi. Die Arbeit nennt 23 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The aim of this study is to develop and investigate a hierarchical system for forecasting and interpreting deformations of hydraulic structures based on geodetic time series of GNSS and TPS observations using machine learning methods. The proposed approach is designed not only to improve the accuracy of coordinate displacement forecasting, but also to form an interpretable monitoring system capable of accounting for temperature effects, missing observations, irregular time intervals, measurement uncertainty, and the different functional roles of sensor subsystems. The study uses geodetic observation time series represented by the coordinates of control points, measurement accuracy indicators, and temperature parameters of the water environment in the area of the hydraulic structure at corresponding depths. For deformation forecasting, a multi-horizon task was formulated to estimate coordinate increments for horizons of 1, 3, 7, and 14 days. GRU and TCN sequence architectures, as well as their ensemble, were used as the main models. To account for incomplete and irregular observations, missingness masks and time intervals between measurements were included in the input representation. In addition, temperature-aware forecasting with component analysis was implemented, enabling the estimation of temperature-associated and residual components of the deformation signal. The reliability of the results was further assessed by estimating and calibrating forecast uncertainty. The obtained results showed that sequence models provide stable improvement compared with the tabular approach and the persistence forecast. The best overall results were obtained using the GRU and TCN ensemble. It was established that the temperature-associated component forms a noticeable part of the predicted signal, especially at medium and longer forecasting horizons. Uncertainty calibration improved the agreement between the estimated uncertainty and the actual residual errors. The hierarchical station analysis showed that GNSS should be used as the primary predictive subsystem, whereas TPS should perform an auxiliary and contextual role in detecting statistically atypical residual deviations. The proposed approach improves the interpretability of geodetic monitoring and provides a basis for developing more justified decision-support systems for early warning in hydraulic structure monitoring.

Zitieren

APA

Hrabovyi, O. (2026). Hierarchical system for forecasting and decomposition analysis of hydraulic structure deformations based on geodetic time series. Geodesy, cartography and aerial photography, 103,2026(103), 66-80. https://doi.org/10.23939/istcgcap2026.103.066

Vancouver

Hrabovyi O. Hierarchical system for forecasting and decomposition analysis of hydraulic structure deformations based on geodetic time series. Geodesy, cartography and aerial photography. 2026;103,2026(103):66-80. doi:10.23939/istcgcap2026.103.066

BibTeX
@article{Hrabovyi2026,
  author = {Hrabovyi, Oleksandr},
  title = {Hierarchical system for forecasting and decomposition analysis of hydraulic structure deformations based on geodetic time series},
  journal = {Geodesy, cartography and aerial photography},
  year = {2026},
  volume = {103,2026},
  number = {103},
  pages = {66--80},
  doi = {10.23939/istcgcap2026.103.066}
}
Kostenlos starten →