基于数据-知识驱动的高精度海底地形绘制:以南海为例

doi:10.3969/j.issn.1001-909X.2024.03.012

海洋学研究 ›› 2024, Vol. 42 ›› Issue (3): 142-152.DOI: 10.3969/j.issn.1001-909X.2024.03.012

• 研究论文 • 上一篇

基于数据-知识驱动的高精度海底地形绘制:以南海为例

刘洋¹^,²(), 李三忠¹^,²^,^*(), 邹卓延¹^,², 索艳慧¹^,², 孙毅¹^,²

1.中国海洋大学海洋地球科学学院,深海圈层与地球系统教育部前沿科学中心,海底科学与探测技术教育部重点实验室,山东青岛 266100
2.崂山实验室海洋矿产资源评价与探测技术功能实验室,山东青岛 266237

收稿日期:2023-09-27 修回日期:2024-05-22 出版日期:2024-09-15 发布日期:2024-11-25
通讯作者: ^*李三忠(1968—),男,教授,主要从事海底科学与地球系统动力学研究,E-mail:sanzhong@ouc.edu.cn。
作者简介:刘洋(1997—),男,湖北省阳新县人,主要从事海洋地质方面的研究,E-mail:liuyang9553@stu.ouc.edu.cn。
基金资助:
青岛海洋科学与技术试点国家实验室山东省专项经费项目(2022QNLM05032);国家自然科学基金项目(42121005);国家自然科学基金项目(91958214);山东省自然科学基金项目(ZR2021ZD09);111项目(B20048)

High-precision seafloor topographic mapping based on data-knowledge-driven: An example from the South China Sea

LIU Yang¹^,²(), LI Sanzhong¹^,²^,^*(), ZOU Zhuoyan¹^,², SUO Yanhui¹^,², SUN Yi¹^,²

1. Frontiers Science Center for Deep Ocean Multispheres and Earth System, Key Lab of Submarine Geosciences and Prospecting Techniques, MOE and College of Marine Geosciences, Ocean University of China, Qingdao 266100, China
2. Laboratory for Marine Mineral Resources, Laoshan Laboratory, Qingdao 266237, China

Received:2023-09-27 Revised:2024-05-22 Online:2024-09-15 Published:2024-11-25

摘要/Abstract

摘要：

海底地形具有非常重要的商业、工程、军事和科学研究价值。目前,常用重力场数据反演海底地形,如自由空气重力异常和垂直重力梯度。然而,由于现有方法反演海底地形具有较强的多解性,仍然无法准确获取高精度的海底地形。该文提出了重力-密度法与随机森林结合的数据-知识驱动新方法,以重建准确的海底地形。该方法在中国南海海域进行了测试,并与重力-密度法、随机森林以及现有的SIO模型进行了对比分析。反演结果显示,数据-知识驱动提供了更好的反演性能,随机森林和重力-密度法次之,SIO模型最差。相比于重力-密度法,数据-知识驱动的平均绝对误差、平均相对误差和均方根误差分别降低了21%、25%和7%;而相比于随机森林,它们分别也降低了20%、20%和20%。此外,数据-知识驱动模型与船载测深数据具有较高的一致性,其差值大约有72%分布在±10 m范围内,占比高于其他三种模型。该结果证明了数据-知识驱动方法在海底地形反演中的可行性和有效性,有助于加快高精度海底地形的绘制。

关键词: 海底地形, 机器学习, 数据驱动, 知识驱动, 重力-密度法, 随机森林, SIO模型, 船载测深

Abstract:

Seafloor topography is of considerable value in commercial, engineering, military and scientific research. Currently, gravity field data, such as free air gravity anomalies and vertical gravity gradients, are commonly used to inverse seafloor topography. However, due to the strong multi-resolution of the existing methods to inverse seafloor topography, it is still impossible to obtain accurate high-precision seafloor topography. A new data-knowledge-driven method was proposed to reconstruct accurate seafloor topography, which combines the gravity-density method with random forests. This method was applied to the South China Sea and compared with the gravity-density, random forest, and existing SIO models. The inversion results show that the data-knowledge-driven method provides better inversion performance, followed by the random forest and gravity-density methods, and the SIO model is the worst. The mean absolute error, mean relative error and root mean square error of the data-knowledge-driven are reduced by 21%, 25% and 7%, respectively, compared to those of the gravity-density method, while they are also reduced by 20%, 20% and 20%, respectively, compared to those of the random forest. In addition, the data-knowledge-driven model has a high degree of agreement with the shipboard bathymetry data, with approximately 72% of its differences distributed within ±10 m, which is higher than that of the other three models. The results demonstrate the feasibility and effectiveness of the data-knowledge-driven approach in seafloor topography inversion, which helps to accelerate the mapping of high-precision seafloor topography.

Key words: seafloor topography, machine learning, data-driven, knowledge-driven, gravity-density method, random forest, SIO model, shipboard bathymetry

中图分类号:

P714.7

刘洋, 李三忠, 邹卓延, 索艳慧, 孙毅. 基于数据-知识驱动的高精度海底地形绘制:以南海为例[J]. 海洋学研究, 2024, 42(3): 142-152.

LIU Yang, LI Sanzhong, ZOU Zhuoyan, SUO Yanhui, SUN Yi. High-precision seafloor topographic mapping based on data-knowledge-driven: An example from the South China Sea[J]. Journal of Marine Sciences, 2024, 42(3): 142-152.

图/表 10

图1 船载测深数据分布 (蓝色点和橙色点分别代表控制点和检查点,背景为SIOv25.1的水深模型。)

Fig.1 Distribution of shipboard bathymetric data (The blue and orange points represent control and check points, respec-tively, with the bathymetric model of SIOv25.1 in the background.)

图2 研究区的地球物理数据分布

Fig.2 Distribution of geophysical data in the study area

表1 研究区地球物理数据的统计分析

Tab.1 Statistical analysis of geophysical data in the study area

特征参数	最大值	最小值	平均值	标准差
自由空气重力异常/mGal	207.90	-60.64	4.87	19.66
垂直重力梯度/E	281.23	-114.34	0.34	19.97
均衡重力异常/mGal	80.38	-97.52	14.13	18.56
磁异常/nT	217.55	-172.32	5.95	43.39

图3 重力-密度法的原理示意图

Fig.3 Schematic diagram of the principle of gravity-density method

图4 不同密度差异常数下重力-密度法与船载测深之间的相关系数和标准差

Fig.4 Correlation coefficients and standard deviations between gravity-density method and shipboard bathymetry for different density difference constants

表2 基于两种驱动方式的随机森林的最优参数

Tab.2 Optimal parameters for random forests based on two driving methods

参数	最优值
参数	数据驱动	数据-知识驱动
n_eatimators	1 000	1 000
max_depth	60	50
max_features	4	3
min_samples_leaf	1	2
min_samples_split	2	2

图5 四种模型反演的研究区海底地形

Fig.5 Seafloor topography of the study area by four models

表3 四种模型的评估指标

Tab.3 Evaluation indexes for the four models

评估指标	SIO	重力-密度法	随机森林	数据-知识驱动
平均绝对误差/m	83.18±138.66	24.86±51.85	24.51±62.02	19.65±49.59
平均相对误差/%	3.47±7.17	1.09±4.07	1.03±4.09	0.82±3.81
均方根误差/m	161.69	57.50	66.68	53.34
相关系数	0.990 5	0.998 8	0.998 4	0.999 0

图6 船载测深与模型反演深度的比较

Fig.6 Comparison of shipboard bathymetry and inversion depths by four models

图7 四种模型的反演结果与船载测深的差值统计直方图

Fig.7 Histogram of the difference statistics among the inversion results of the four models and shipboard bathymetry

参考文献 47

[1]	WÖLFL A C, SNAITH H, AMIREBRAHIMI S, et al. Seafloor mapping-the challenge of a truly global ocean bathymetry[J]. Frontiers in Marine Science, 2019, 6: 283.
[2]	SANDWELL D T, SMITH W H, GILLE S, et al. Bathymetry from space: White paper in support of a high-resolution, ocean altimeter mission[J]. International Geophysics Series, 2001, 69: 1049-1062.
[3]	SMITH W. Introduction to this special issue on bathymetry from space[J]. Oceanography, 2004, 17(1): 6-7.
[4]	PEERI S, GARDNER J V, WARD L G, et al. The seafloor: A key factor in lidar bottom detection[J]. IEEE Transactions on Geoscience and Remote Sensing, 2011, 49(3): 1150-1157.
[5]	EUGENIO F, MARCELLO J, MARTIN J. High-resolution maps of bathymetry and benthic habitats in shallow-water environments using multispectral remote sensing imagery[J]. IEEE Transactions on Geoscience and Remote Sensing, 2015, 53(7): 3539-3549.
[6]	黄谟涛, 翟国君, 欧阳永忠, 等. 卫星测高技术应用研究回顾与展望[J]. 海洋测绘, 2004, 24(4):65-70.
	HUANG M T, ZHAI G J, OU-YANG Y Z, et al. Review and prospect of application study in altimetry[J]. Hydrographic Surveying and Charting, 2004, 24(4): 65-70.
[7]	CALMANT S, BERGE-NGUYEN M, CAZENAVE A. Global seafloor topography from a least-squares inversion of altimetry-based high-resolution mean sea surface and shipboard soundings[J]. Geophysical Journal International, 2002, 151(3): 795-808.
[8]	TOZER B, SANDWELL D T, SMITH W H F, et al. Global bathymetry and topography at 15Arc Sec: SRTM15+[J]. Earth and Space Science, 2019, 6(10): 1847-1864.
[9]	AMANTE C, EAKINS B W. ETOPO1 arc-minute global relief model: Procedures, data sources and analysis[R]//NOAA Technical Memorandum NESDIS NGDC-24. NOAA, National Geophysical Data Center, 2009.
[10]	MARKS K M, SMITH W H F, SANDWELL D T. Evolution of errors in the altimetric bathymetry model used by Google Earth and GEBCO[J]. Marine Geophysical Researches, 2010, 31(3): 223-238.
[11]	SMITH W H F, SANDWELL D T. Global sea floor topography from satellite altimetry and ship depth soundings[J]. Science, 1997, 277(5334): 1956-1962.
[12]	ANDERSEN O B, KNUDSEN P. The DNSC08BAT bathy-metry developed from satellite altimetry[C]// Proceedings of the EGU-2008 Meeting: Vienna, Austria, 2008.
[13]	BECKER J J, SANDWELL D T, SMITH W H F, et al. Global bathymetry and elevation data at 30 arc seconds resolution: SRTM30_PLUS[J]. Marine Geodesy, 2009, 32(4): 355-371.
[14]	胡敏章, 张胜军, 金涛勇, 等. 新一代全球海底地形模型BAT_WHU2020[J]. 测绘学报, 2020, 49(8):939-954. DOI
	HU M Z, ZHANG S J, JIN T Y, et al. A new generation of global bathymetry model BAT_WHU2020[J]. Acta Geodaetica et Cartographica Sinica, 2020, 49(8): 939-954. DOI
[15]	HSIAO Y S, KIM J W, KIM K B, et al. Bathymetry estimation using the gravity-geologic method: An investi-gation of density contrast predicted by the downward continuation method[J]. Terrestrial, Atmospheric and Oceanic Sciences, 2011, 22(3): 347.
[16]	KIM J W, VON FRESE R R B, LEE B Y, et al. Altimetry-derived gravity predictions of bathymetry by the gravity-geologic method[J]. Pure and Applied Geophysics, 2011, 168(5): 815-826.
[17]	欧阳明达. 利用海洋重力数据反演海底地形的理论与方法[D]. 郑州: 解放军信息工程大学, 2015.
	OU-YANG M D. Theory and method of inversion of seabed topography using marine gravity data[D]. Zhengzhou: PLA Information Engineering University, 2015.
[18]	MCNUTT M. Compensation of oceanic topography: An application of the response function technique to the surveyor area[J]. Journal of Geophysical Research: Solid Earth, 1979, 84(B13): 7589-7598.
[19]	ARABELOS D, TZIAVOS I N. Gravity-field improvement in the Mediterranean Sea by estimating the bottom topography using collocation[J]. Journal of Geodesy, 1998, 72(3): 136-143.
[20]	KIM K B, HSIAO Y S, KIM J W, et al. Bathymetry enhancement by altimetry-derived gravity anomalies in the East Sea (Sea of Japan)[J]. Marine Geophysical Researches, 2010, 31(4): 285-298.
[21]	李倩倩, 鲍李峰. 测高重力场反演海底地形方法比较[J]. 海洋测绘, 2016, 36(5):1-4,18.
	LI Q Q, BAO L F. Comparative analysis of methods for bathymetry prediction from altimeter-derived gravity anomalies[J]. Hydrographic Surveying and Charting, 2016, 36(5): 1-4, 18.
[22]	胡敏章, 李建成, 金涛勇. 应用重力地质方法反演皇帝海山的海底地形[J]. 武汉大学学报:信息科学版, 2012, 37(5):610-612,629.
	HU M Z, LI J C, JIN T Y. Bathymetry inversion with gravity-geologic method in Emperor Seamount[J]. Geomatics and Information Science of Wuhan University, 2012, 37(5): 610-612, 629.
[23]	王永康, 周兴华, 唐秋华, 等. 应用重力地质法反演马里亚纳海沟地形[J]. 海洋科学进展, 2020, 38(4):708-716.
	WANG Y K, ZHOU X H, TANG Q H, et al. Predicting bathymetry in Mariana trench using gravity-geologic method[J]. Advances in Marine Science, 2020, 38(4): 708-716.
[24]	欧阳明达, 孙中苗, 翟振和. 基于重力地质法的南中国海海底地形反演[J]. 地球物理学报, 2014, 57(9):2756-2765. DOI
	OU-YANG M D, SUN Z M, ZHAI Z H. Predicting bathymetry in South China Sea using the gravity-geologic method[J]. Chinese Journal of Geophysics, 2014, 57(9): 2756-2765.
[25]	ANNAN R F, WAN X Y. Recovering bathymetry of the gulf of Guinea using altimetry-derived gravity field products combined via convolutional neural network[J]. Surveys in Geophysics, 2022, 43(5): 1541-1561.
[26]	SUN Y J, ZHENG W, LI Z W, et al. Improved the accuracy of seafloor topography from altimetry-derived gravity by the topography constraint factor weight optimization method[J]. Remote Sensing, 2021, 13(12): 2277.
[27]	AN D C, GUO J Y, LI Z, et al. Improved gravity-geologic method reliably removing the long-wavelength gravity effect of regional seafloor topography: A case of bathymetric prediction in the South China Sea[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60: 4211912.
[28]	CALMANT S, BAUDRY N. Modelling bathymetry by inverting satellite altimetry data: A review[J]. Marine Geophysical Researches, 1996, 18(2): 123-134.
[29]	WANG Y M. Predicting bathymetry from the earth’s gravity gradient anomalies[J]. Marine Geodesy, 2000, 23(4): 251-258.
[30]	REICHSTEIN M, CAMPS-VALLS G, STEVENS B, et al. Deep learning and process understanding for data-driven earth system science[J]. Nature, 2019, 566(7743): 195-204.
[31]	SHEN C P, APPLING A, GENTINE P, et al. Differentiable modelling to unify machine learning and physical models for geosciences[J]. Nature Reviews Earth & Environment, 2023, 4: 552-567.
[32]	YANG L, LIU M, LIU N, et al. Recovering bathymetry from satellite altimetry-derived gravity by fully connected deep neural network[J]. IEEE Geoscience and Remote Sensing Letters, 2023, 20: 1502805.
[33]	CHEN Y, ZHANG D. Integration of knowledge and data in machine learning[Z/OL]. arXiv, 2022: 2202. 10337. https://arxiv.org/abs/2202.10337.
[34]	KARPATNE A, ATLURI G, FAGHMOUS J H, et al. Theory-guided data science: A new paradigm for scientific discovery from data[J]. IEEE Transactions on Knowledge and Data Engineering, 2017, 29(10): 2318-2331.
[35]	GETTELMAN A, GEER A J, FORBES R M, et al. The future of earth system prediction: Advances in model-data fusion[J]. Science Advances, 2022, 8(14): eabn3488.
[36]	MENG C Z, SEO S, CAO D F, et al. When physics meets machine learning: A survey of physics-informed machine learning[Z/OL]. arXiv, 2022: 2203. 16797. https://arxiv.org/abs/2203.16797v1.
[37]	MORTON B, BLACKMORE G. South China Sea[J]. Marine Pollution Bulletin, 2001, 42(12): 1236-1263. PMID
[38]	WANG A M, DU Y, PENG S Q, et al. Deep water characteristics and circulation in the South China Sea[J]. Deep Sea Research Part I: Oceanographic Research Papers, 2018, 134: 55-63.
[39]	SANDWELL D T, MÜLLER R D, SMITH W H F, et al. Marine geophysics. New global marine gravity model from CryoSat-2 and Jason-1 reveals buried tectonic structure[J]. Science, 2014, 346(6205): 65-67. DOI PMID
[40]	BALMINO G, VALES N, BONVALOT S, et al. Spherical harmonic modelling to ultra-high degree of Bouguer and isostatic anomalies[J]. Journal of Geodesy, 2012, 86(7): 499-520.
[41]	MAUS S, BARCKHAUSEN U, BERKENBOSCH H, et al. EMAG2: A 2-arc min resolution earth magnetic anomaly grid compiled from satellite, airborne, and marine magnetic measurements[J]. Geochemistry, Geophysics, Geosystems, 2009, 10(8). https://doi.org/10.1029/2009GC002471.
[42]	PARKER R L. The rapid calculation of potential anomalies[J]. Geophysical Journal International, 1973, 31(4): 447-455.
[43]	HWANG C. A bathymetric model for the South China sea from satellite altimetry and depth data[J]. Marine Geodesy, 1999, 22(1): 37-51.
[44]	KIM K B, YUN H S. Satellite-derived bathymetry prediction in shallow waters using the gravity-geologic method: A case study in the west sea of Korea[J]. KSCE Journal of Civil Engineering, 2018, 22(7): 2560-2568.
[45]	HSIAO Y S, HWANG C, CHENG Y S, et al. High-resolution depth and coastline over major atolls of South China Sea from satellite altimetry and imagery[J]. Remote Sensing of Environment, 2016, 176: 69-83.
[46]	STRYKOWSKI G, BOSCHETTI F, PAPP G. Estimation of the mass density contrasts and the 3D geometrical shape of the source bodies in the Yilgarn area, Eastern Goldfields, Western Australia[J]. Journal of Geodynamics, 2005, 39: 444-460.
[47]	BREIMAN L. Random forests[J]. Machine Learning, 2001, 45: 5-32.

基于数据-知识驱动的高精度海底地形绘制:以南海为例

High-precision seafloor topographic mapping based on data-knowledge-driven: An example from the South China Sea

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 10

参考文献 47

相关文章 6

编辑推荐

Metrics

[1]	陆钰婷, 郭文康, 丁骏, 王林峰, 李晓辉, 王久珂. 人工智能海浪预报的发展与挑战[J]. 海洋学研究, 2024, 42(3): 28-37.
[2]	金阳, 韩磊, 金梅兵, 董昌明. 基于ConvLSTM的中国东南沿海波浪智能预报和评估[J]. 海洋学研究, 2024, 42(3): 88-98.
[3]	罗通, 洪加诚. 基于机器学习的热带气旋快速增强预报[J]. 海洋学研究, 2024, 42(3): 99-107.
[4]	游高冲, 郭杭, 罗孝文, 等. 基于LS-SVM的多系统融合GNSS-MR潮位反演[J]. 海洋学研究, 2022, 40(1): 72-.
[5]	张莹, 谭艳春, 彭发定, 廖杏杰, 余昱昕. 基于EEMD和ARIMA的海温预测模型研究[J]. 海洋学研究, 2019, 37(1): 9-14.
[6]	姚才华, 吴自银. 30a来伶仃洋海岸线变迁及海底冲淤变化[J]. 海洋学研究, 2012, 30(3): 89-100.