Spatial-temporal evolution characteristics and prediction of land subsidence in the eastern plain of Beijing
YU Wen1,2,3,4(), GONG Huili1,2,3,4(), CHEN Beibei1,2,3,4, ZHOU Chaofan1,2,3,4
1. Beijing Laboratory of Water Resources Security, Capital Normal University, Beijing 100048, China 2. Key Laboratory of Land Subsidence Mechanism and Control, Ministry of Education, Capital Normal University, Beijing 100048, China 3. Base of the State Key Laboratory of Urban Environmental Process and Digital Modeling, Capital Normal University, Beijing 100048, China 4. National Field Scientific Observation and Research Station of Groundwater and Land Subsidence in the Beijing-Tianjin-Hebei Plain, Capital Normal University, Beijing 100048, China
Land subsidence is a natural geological phenomenon in which the surface elevation drops. It can severely destroy urban infrastructure and threaten urban safety if it occurs in densely populated cities with a high social development degree. The analysis of the evolution characteristics of land subsidence can reflect the degree of the influence of land subsidence on the ground infrastructures, and building an efficient land subsidence prediction model is of great significance for preventing and controlling land subsidence and protecting urban safety. This study obtained the spatial-temporal information on land subsidence using the persistent scatterer interferometric synthetic aperture Radar (PS-InSAR) method first and then verified the information using leveling to get high precision. Then, this study analyzed the general spatial-temporal characteristics of the land subsidence field using an empirical orthogonal function. The analysis results are as follows. Spatial modal No. 1 had a high variance contribution rate, almost representing the general spatial evolution of the study area. Its corresponding time coefficient showed a significant linear trend. By contrast, spatial mode No. 2 had a low variance contribution rate and a seasonally significant time coefficient. Finally, the time series of the regional land subsidence were predicted using both long short-term memory (LSTM) and Attention-LSTM models. The prediction results indicate that the Attention-LSTM model was superior to the LSTM model, with the mean square error loss (MSE-loss) of as low as 0.01. This prediction method expands the application of deep learning in the study of land subsidence.
Wen YU,Huili GONG,Beibei CHEN, et al. Spatial-temporal evolution characteristics and prediction of land subsidence in the eastern plain of Beijing[J]. Remote Sensing for Natural Resources,
2022, 34(4): 183-193.
Fig.3 Land subsidence prediction framework using Attention-LSTM
Fig.4 Cumulative settlement of the study area
Fig.5 Time series annual settlement map of the study area
Fig.6 Verification of InSAR results and leveling results
Fig.7 Distribution of subsidence fecture vectors and corresponding time coefficients of typical subsidence areas in the eastern plain of Beijing from 2011 to 2018
Fig.8 Distribution of monthly-scale subsidence feature vectors and corresponding time coefficients of typical subsidence area in the eastern plain of Beijing from 2012 to 2014
Fig.9 Distribution of monthly-scale subsidence feature vectors and corresponding time coefficients of typical subsidence area in the eastern plain of Beijing from 2016 to 2018
Fig.10 Forecast model loss function
Fig.11 Regional settlement prediction results of LSTM and Attention-LSTM
Fig.12 Comparison between the real value and the predicted value of the selected profile in the study area
Zhou F F. Interpretation of the national plan for land subsidence prevention and control (2011—2020):Interview with Tao Qingfa,deputy director of the department of geological environment,Ministry of Land and Resources[J]. China Emergency Management, 2012(3):58-61.
[2]
叶晓宾. 华北平原地面沉降经济损失评估[M]. 北京: 中国大地出版社, 2006.
[2]
Ye X B. Evaluation of economic loss of land subsidence in the North China Plain[M]. Beijing: China Land Publishing House, 2006.
Liu G X. Application examples of InSAR and its limitation analysis[J]. Surveying and Mapping of Sichuan, 2005, 28(3):139-143.
[4]
Zebker H A, Goldstein R M. Topographic mapping from interferometric synthetic aperture Radar observations[J]. Journal of Geophysical Research Solid Earth, 1986, 91(b5):4993-4999.
[5]
Massonnet D, Rossi M, Carmona C, et al. The displacement field of the Landers earthquake mapped by Radar interferometry[J]. Nature, 1993, 364(6433):138-142.
doi: 10.1038/364138a0
url: https://doi.org/10.1038/364138a0
Gong H L, Zhang Y Q, Li X J. Beijing land subsidence research based on permanent scatterer Radar interferometry technology[J]. Advances in Natural Science, 2009, 19(11):1261-1266.
[7]
Gabriel A K, Goldstein R M, Zebker H A. Mapping small elevation changes over large areas:Differential Radar interferometry[J]. Journal of Geophysical Research:Solid Earth, 1989, 94(b7):9183-9191.
[8]
Hanssen R F. Radar interferometry[M]. Springer Netherlands, 2001.
[9]
Ghiglia D C, Pritt M D. Two-dimensional phase unwrapping[M]. Wiley-Interscience, 1985.
[10]
Zebker H A, Rosen P A, Hensley S. Atmospheric effects in interfero-metric synthetic aperture Radar surface deformation and topographic maps[J]. Journal of Geophysical Research Solid Earth, 1997, 102(b4):7547-7563.
[11]
Ferretti A, Prati C. Nonlinear subsidence rate estimation using permanent scatterers in differential SAR interferometry[J]. IEEE Transactions on Geoscience and Remote Sensing, 2000, 38(5):2202-2212.
doi: 10.1109/36.868878
url: http://ieeexplore.ieee.org/document/868878/
Yang C Y, Wang Y B, Zhao Y L, et al. Temporal and spatial characteristics of land subsidence in Laiguangying,Beijing[J]. Remote Sensing Information, 2020, 35(5):138-143.
Yang T L, Xu Y. Research trends in international land subsidence and urban security:An overview of the first international symposium on urban geology[J]. Shanghai Land and Resources, 2017, 38(2):1-3.
Liu Q H, Zhang Y H, Deng M, et al. Time series prediction method of large-scale surface subsidence based on deep learning[J]. Acta Geodaetica et Cartographica Sinica, 2021, 50(3):396-404.
Yue Z H, Shen T, Mao X, et al. Ground subsidence prediction method based on recurrent neural network[J]. Science of Surveying and Mapping, 2020, 45(12):149-156.
Liu Y. Divided water-bearing zones and compressible zones of Beijing land subsidence area and land subsidence automatic monitoring system[D]. Changchun: Jilin University, 2004.
Zhou Y, Luo Y, Guo G X, et al. A study of the characteristics of land subsidence and the main control factors in the alluvial plain:A case study of Beijing Plain[J]. Geological Bulletin of China, 2016, 35(12):2100-2110.
Cheng L P, Wang X H, Zhang Q W, et al. Influence of transferring Yangtze River water into Beijing on ground subsidence and trend analysis[J]. Yellow River, 2018, 40(5):93-97.
Liu Y Y. Research on time series InSAR surface deformation monitoring based on multi-source SAR Data[D]. Xi’an: Chang’an University, 2014.
[25]
Lorenz E N. Empirical orthogonal functions and statistical weather prediction[M]. Cambridge: Massachusetts Institute of Technology, Department of Meteorology, 1956:1,52.
[26]
Asoka A, Gleeson T, Wada Y, et al. Relative contribution of monsoon precipitation and pumping to changes in groundwater storage in India[J]. Nature Geoscience, 2017, 10(2):109-117.
doi: 10.1038/ngeo2869
url: https://doi.org/10.1038/ngeo2869
Bianchi F M, Maiorino E, Kampffmeyer M C, et al. Recurrent neural networks for short-term load forecasting:An overview and comparative analysis[EB/OL].(2017-05-11) [2018-07-20] https://arxiv.org/pdf/1705.04378.pdf .
url: https://arxiv.org/pdf/1705.04378.pdf