利用机器学习模型预测中国沿海海平面变化

doi:10.3969/j.issn.1001-909X.2024.03.009

摘要/Abstract

摘要：

该文利用线性回归函数,根据卫星测高及中国沿海6个验潮站数据估算出1993—2020年中国沿海绝对海平面上升速率为4.17±1.32 mm/a,相对海平面上升速率为4.47±0.90 mm/a。将1958—2020年的大气数据、海洋数据及气候模态指数作为预报因子,建立了长短期记忆神经网络模型(LSTM模型)、循环神经网络模型(RNN模型)、门控循环单元神经网络模型(GRU模型)和支持向量机回归模型(SVR模型)等多种神经网络模型对中国沿海6个验潮站周边的相对海平面变化趋势进行预测。模型评估结果表明,同时引入大气变量、海洋变量及气候模态指数变量的LSTM模型取得的预测值与观测值的平均相关系数和均方根误差分别为0.866和19.279 mm,在4种模型中表现最佳,可以作为一种新型的预测相对海平面变化的方法。

关键词: 海平面上升, 相对海平面, 绝对海平面, 垂直地壳运动, 卫星测高, 潮位, LSTM神经网络模型, 时间序列预测

Abstract:

Based on the data of satellite altimetry and six tide gauge stations along the coast of China, linear regression function was used to estimate the absolute sea level rise rate in the coastal areas of China from 1993 to 2020, which was 4.17±1.32 mm/a, and the relative sea level rise rate was 4.47±0.90 mm/a. Taking the atmospheric data, ocean data and climate modal index from 1958 to 2020 as prediction factors, a variety of neural network models such as long short-term memory neural network model (LSTM model), recurrent neural network model (RNN model), gated recurrent unit neural network model (GRU model) and support vector machine regression model (SVR model) were established to predict the trend of relative sea level changes around the six tide gauge stations along the coast of China. The model evaluation results show that the average correlation coefficient and root mean square error of the observed value and the predicted value obtained by the LSTM model that simultaneously introduces atmospheric and ocean variables and climate modal index variables are 0.866 and 19.279 mm, respectively, which performs the best among the four models, and therefore the LSTM model can be used as a new method for predicting relative sea level changes.

Key words: sea level rise, relative sea level, absolute sea level, vertical land movement, satellite altimetry, tide level, LSTM neural network model, time series forecasting

中图分类号:

P731.34

陈建珩, 许东峰, 姚志雄. 利用机器学习模型预测中国沿海海平面变化[J]. 海洋学研究, 2024, 42(3): 108-118.

CHEN Jianheng, XU Dongfeng, YAO Zhixiong. Prediction of sea level changes along the coast of China using machine learning models[J]. Journal of Marine Sciences, 2024, 42(3): 108-118.

图/表 13

参考文献 50

[1]	MASSON-DELMOTTE V, ZHAI P, PIRANI A, et al. Climate change 2021: The physical science basis[M]. Cambridge, UK and New York, NY, USA: Cambridge University Press, 2021: 2391.
[2]	VERMEER M, RAHMSTORF S. Global sea level linked to global temperature[J]. Proceedings of the National Academy of Sciences of the United States of America, 2009, 106(51): 21527-21532. DOI PMID
[3]	OPPENHEIMER M, GLAVOVIC B, HINKEL J, et al. Sea level rise and implications for low-lying islands, coasts and communities[M]//IPCC special report on the ocean and cryosphere in a changing climate. Cambridge, UK and New York, NY, USA: Cambridge University Press, 2019: 321-445.
[4]	NEREM R S, BECKLEY B D, FASULLO J T, et al. Climate-change-driven accelerated sea-level rise detected in the altimeter era[J]. Proceedings of the National Academy of Sciences of the United States of America, 2018, 115(9): 2022-2025. DOI PMID
[5]	自然资源部. 2022年中国海平面公报[EB/OL].(2023-04-05)[2023-08-25]. http://gi.mnr.gov.cn/202304/t20230412_2781114.html.
[6]	WOODWORTH P L, MELET A, MARCOS M, et al. Forcing factors affecting sea level changes at the coast[J]. Surveys in Geophysics, 2019, 40(6): 1351-1397. DOI
[7]	WANG Q S, PAN C H, ZHANG G Z. Impact of and adaptation strategies for sea-level rise on Yangtze River Delta[J]. Advances in Climate Change Research, 2018, 9(2): 154-160.
[8]	FENG X B, CHENG Y C. Observing sea levels in the China Seas from satellite altimetry[M]//Remote sensing of the Asian seas. Cham: Springer International Publishing, 2018: 321-338.
[9]	MENG L S, ZHUANG W, ZHANG W W, et al. Decadal sea level variability in the Pacific Ocean: Origins and climate mode contributions[J]. Journal of Atmospheric and Oceanic Technology, 2019, 36(4): 689-698.
[10]	XI H, ZHANG Z Z, LU Y, et al. Long-term and interannual variation of the steric sea level in the South China Sea and the connection with ENSO[J]. Journal of Coastal Research, 2019, 35(3): 489.
[11]	SITHARA S, PRAMADA S K, THAMPI S G. Sea level prediction using climatic variables: A comparative study of SVM and hybrid wavelet SVM approaches[J]. Acta Geophysica, 2020, 68(6): 1779-1790.
[12]	ZUBIER K M, EYOUNI L S. Investigating the role of atmospheric variables on sea level variations in the eastern central Red Sea using an artificial neural network approach[J]. Oceanologia, 2020, 62(3): 267-290.
[13]	BALOGUN A L, ADEBISI N. Sea level prediction using ARIMA, SVR and LSTM neural network: Assessing the impact of ensemble ocean-atmospheric processes on models’ accuracy[J]. Geomatics, Natural Hazards and Risk, 2021, 12(1): 653-674.
[14]	ADEBISI N, BALOGUN A L. A deep-learning model for national scale modelling and mapping of sea level rise in Malaysia: The past, present, and future[J]. Geocarto International, 2022, 37(23): 6892-6914.
[15]	MUSLIM T O, AHMED A N, MALEK M A, et al. Investigating the influence of meteorological parameters on the accuracy of sea-level prediction models in Sabah, Malaysia[J]. Sustainability, 2020, 12(3): 1193.
[16]	FU Y G, ZHOU X H, SUN W K, et al. Hybrid model combining empirical mode decomposition, singular spectrum analysis, and least squares for satellite-derived sea-level anomaly prediction[J]. International Journal of Remote Sensing, 2019, 40(20): 7817-7829.
[17]	ISLAM B U, AHMED S F. Short-term electrical load demand forecasting based on LSTM and RNN deep neural networks[J]. Mathematical Problems in Engineering, 2022, 2022: 2316474.
[18]	MAKRIDAKIS S, SPILIOTIS E, ASSIMAKOPOULOS V. Statistical and machine learning forecasting methods: Concerns and ways forward[J]. PLoS One, 2018, 13(3): e0194889.
[19]	FARINOTTI D, HUSS M, FÜRST J J, et al. A consensus estimate for the ice thickness distribution of all glaciers on earth[J]. Nature Geoscience, 2019, 12: 168-173.
[20]	SAENKO O A, YANG D, GREGORY J M, et al. Separating the influence of projected changes in air temperature and wind on patterns of sea level change and ocean heat content[J]. Journal of Geophysical Research: Oceans, 2015, 120(8): 5749-5765.
[21]	SMITH-KONTER B R, THORNTON G M, SANDWELL D T. Vertical crustal displacement due to interseismic deformation along the San Andreas fault: Constraints from tide gauges[J]. Geophysical Research Letters, 2014, 41(11): 3793-3801.
[22]	SUZUKI T, ISHII M. Long-term regional sea level changes due to variations in water mass density during the period 1981-2007[J]. Geophysical Research Letters, 2011, 38(21): L21604.
[23]	OSTANCIAUX É, HUSSON L, CHOBLET G, et al. Present-day trends of vertical ground motion along the coast lines[J]. Earth-Science Reviews, 2012, 110(1/2/3/4): 74-92.
[24]	PFEFFER J, SPADA G, MÉMIN A, et al. Decoding the origins of vertical land motions observed today at coasts[J]. Geophysical Journal International, 2017, 210(1): 148-165.
[25]	RAUCOULES D, LE COZANNET G, WÖPPELMANN G, et al. High nonlinear urban ground motion in Manila (Philippines) from 1993 to 2010 observed by DInSAR: Implications for sea-level measurement[J]. Remote Sensing of Environment, 2013, 139: 386-397.
[26]	HELIANI L S, WIDJAJANTI N, ENDRAYANTO I, et al. Preprocessing of coastal satellite altimetry, tide gauges, and GNSS data: Towards the possibility of detected vertical deformation of south Java Island[J]. Procedia Environ-mental Sciences, 2013, 17: 308-316.
[27]	ROVERE A, RAYMO M E, VACCHI M, et al. The analysis of last interglacial (MIS 5e) relative sea-level indicators: Reconstructing sea-level in a warmer world[J]. Earth-Science Reviews, 2016, 159: 404-427.
[28]	VIGNUDELLI S, BIROL F, BENVENISTE J, et al. Satellite altimetry measurements of sea level in the coastal zone[J]. Surveys in Geophysics, 2019, 40(6): 1319-1349. DOI
[29]	TANG X H, WANG F, CHEN Y L, et al. Warming trend in northern East China Sea in recent four decades[J]. Chinese Journal of Oceanology and Limnology, 2009, 27(2): 185-191.
[30]	CHEN Z Y, WANG Z H. Yangtze Delta, China: Taihu lake-level variation since the 1950s, response to sea-level rise and human impact[J]. Environmental Geology, 1999, 37(4): 333-339.
[31]	WU P, VAN DER WAL W. Postglacial sealevels on a spherical, self-gravitating viscoelastic earth: Effects of lateral viscosity variations in the upper mantle on the inference of viscosity contrasts in the lower mantle[J]. Earth and Planetary Science Letters, 2003, 211(1/2): 57-68.
[32]	YIN J, YU D P, YIN Z E, et al. Modelling the combined impacts of sea-level rise and land subsidence on storm tides induced flooding of the Huangpu River in Shanghai, China[J]. Climatic Change, 2013, 119(3): 919-932.
[33]	ZOU F, TENZER R, FOK H S, et al. The sea-level changes in Hong Kong from tide-gauge records and remote sensing observations over the last seven decades[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2021, 14: 6777-6791.
[34]	郭金运, 王建波, 胡志博, 等. 由TOPEX/Poseidon和Jason-1/2探测的1993—2012中国海海平面时空变化[J]. 地球物理学报, 2015, 58(9):3103-3120. DOI
	GUO J Y, WANG J B, HU Z B, et al. Temporal-spatial variations of sea level over China seas derived from altimeter data of TOPEX/Poseidon, Jason-1 and Jason-2 from 1993 to 2012[J]. Chinese Journal of Geophysics, 2015, 58(9): 3103-3120.
[35]	李建成, 王正涛, 胡建国. 联合多种卫星测高数据分析全球和中国海海平面变化[J]. 武汉测绘科技大学学报, 2000, 25(4):343-347.
	LI J C, WANG Z T, HU J G. Mean sea level variation using historic satellite altimeter data[J]. Geomatics and Information Science of Wuhan University, 2000, 25(4): 343-347.
[36]	CHENG X H, QI Y Q. Trends of sea level variations in the South China Sea from merged altimetry data[J]. Global and Planetary Change, 2007, 57(3/4): 371-382.
[37]	詹金刚, 王勇, 柳林涛. 中国近海海平面季节尺度变化的时频分析[J]. 地球物理学报, 2003, 46(1):36-41.
	ZHAN J G, WANG Y, LIU L T. Time-frequency analysis of the inter-seasonal variations of China-neighboring seas level[J]. Chinese Journal of Geophysics, 2003, 46(1): 36-41.
[38]	AL-JABERY K, OBAFEMI-AJAYI T, OLBRICHT G, et al. Computational learning approaches to data analytics in biomedical applications[M]. London: Academic Press, 2020.
[39]	ISHIDA K, TSUJIMOTO G, ERCAN A L, et al. Hourly-scale coastal sea level modeling in a changing climate using long short-term memory neural network[J]. Science of the Total Environment, 2020, 720: 137613.
[40]	BERGSTRA J, BENGIO Y. Random search for hyper-parameter optimization[J]. Journal of Machine Learning Research, 2012, 13: 281-305.
[41]	WÖPPELMANN G, MARCOS M. Vertical land motion as a key to understanding sea level change and variability[J]. Reviews of Geophysics, 2016, 54(1): 64-92.
[42]	于振龙, 许东峰, 姚志雄, 等. 基于多变量LSTM神经网络模型的PDO指数预测研究[J]. 海洋学报, 2022, 44(6):58-67.
	YU Z L, XU D F, YAO Z X, et al. Research on PDO index prediction based on multivariate LSTM neural network model[J]. Haiyang Xuebao, 2022, 44(6): 58-67.
[43]	HYNDMAN R J, KOEHLER A B. Another look at measures of forecast accuracy[J]. International Journal of Forecasting, 2006, 22(4): 679-688.
[44]	CHEN N, HAN G Q, YANG J S. Mean relative sea level rise along the coasts of the China Seas from mid-20th to 21st centuries[J]. Continental Shelf Research, 2018, 152: 27-34.
[45]	PARKER A. Relative sea level rise along the coast of China mid-twentieth to end twenty-first centuries[J]. Arabian Journal of Geosciences, 2018, 11: 262.
[46]	ZHOU D X, LIU Y, FENG Y K, et al. Absolute sea level changes along the coast of China from tide gauges, GNSS, and satellite altimetry[J]. Journal of Geophysical Research: Oceans, 2022, 127(9): e2022JC018994.
[47]	QU Y, JEVREJEVA S, JACKSON L P, et al. Coastal sea level rise around the China Seas[J]. Global and Planetary Change, 2019, 172: 454-463. DOI
[48]	CONNOR J T, MARTIN R D, ATLAS L E. Recurrent neural networks and robust time series prediction[J]. IEEE Transactions on Neural Networks, 1994, 5(2): 240-254. DOI PMID
[49]	CHUNG J, GULCEHRE C, CHO K H, et al. Empirical evaluation of gated recurrent neural networks on sequence modeling[Z/OL]. arXiv, 2023: 1412.3555. https://arxiv.org/abs/1412.3555.
[50]	CORTES C, VAPNIK V. Support-vector networks[J]. Machine Learning, 1995, 20(3): 273-297.

站点	站点号	起始年份	纬度/(°N)	经度/(°E)	1993—2020年数据完整度/%
大连	723	1954年	38.867	121.683	98.5
吕四	979	1961年	32.133	121.617	87.7
坎门	934	1959年	28.083	121.283	97.6
大埔滘	1034	1963年	22.442	114.184	98.8
闸坡	933	1959年	21.583	111.817	98.8
西沙	1745	1990年	16.833	112.333	98.8

站点	站点号	起始年份	纬度/(°N)	经度/(°E)	1993—2020年数据完整度/%
大连	723	1954年	38.867	121.683	98.5
吕四	979	1961年	32.133	121.617	87.7
坎门	934	1959年	28.083	121.283	97.6
大埔滘	1034	1963年	22.442	114.184	98.8
闸坡	933	1959年	21.583	111.817	98.8
西沙	1745	1990年	16.833	112.333	98.8

数据集划分比例(训练集∶验证集)	归一化RMSE	PCC
5∶5	0.415	0.309
6∶4	0.325	0.460
7∶3	0.214	0.739
8∶2	0.108	0.828
9∶1	0.095	0.839

数据集划分比例(训练集∶验证集)	归一化RMSE	PCC
5∶5	0.415	0.309
6∶4	0.325	0.460
7∶3	0.214	0.739
8∶2	0.108	0.828
9∶1	0.095	0.839

变量组合	变量数量/个	归一化RMSE	PCC
OV	2	0.168	0.788
AV	5	0.259	0.564
IV	2	0.109	0.821
OAV	7	0.264	0.569
OIV	4	0.086	0.843
AIV	7	0.178	0.766
OAIV	9	0.072	0.857