基于SIT-TransNET的被动微波海冰厚度反演

doi:10.19678/j.issn.1000-3428.0070709

摘要/Abstract

摘要： 海冰厚度是全球气候变化研究中的关键参数之一，其在调节地球气候系统、海洋环流和热量交换中具有重要作用。但是由于海冰的物理特性高度变化的影响，使得海冰厚度的精确反演面临巨大挑战。针对上述问题，本文提出了一种多特征融合和改进Transformer的被动微波遥感海冰厚度反演方法SIT-TransNET，该方法利用AMSR2卫星的亮度温度数据，并结合辅助数据（包括雪表面温度、海面盐度和1.4GHz亮度温度），探讨了这些数据与海冰厚度之间的复杂关联，分析了不同特征的重要性并通过建立不同的特征融合方案，加强了对海冰厚度的有效表征；通过SIT-TransNet模型的自注意力机制和多头注意力机制捕捉不同特征及组合对于海冰厚度反演的贡献，并能够动态调整不同特征的权重，实现海冰厚度的精确反演。实验结果表明，相比其他方法，本文提出的SIT-TransNET方法显著提高了海冰厚度反演的精度，决定系数（R²）达到了0.96，均方根误差（RMSE）为9cm，表明此方法适用于海冰厚度反演，为实现大范围海冰厚度监测和气候变化研究提供了有效的技术手段。

Abstract: Sea ice thickness, a key parameter in global climate change research, plays a vital role in regulating the Earth's climate system, ocean circulation, and heat exchange. However, due to the influence of the height change of the physical characteristics of sea ice, the accurate inversion of sea ice thickness faces great challenges. In response to these issues, this paper presents an improved passive microwave remote sensing method for sea ice thickness inversion, SIT-TransNET, which utilizes brightness temperature data from the AMSR2 satellite. The method incorporates auxiliary data, including surface snow temperature, sea surface salinity, and 1.4 GHz brightness temperature, to explore complex correlations with sea ice thickness. Analysis of the importance of different features and the establishment of various feature fusion schemes enhance the effective representation of sea ice thickness. Through the self-attention mechanism and multi-head attention mechanism of the SIT-TransNET model, contributions of different features and their combinations to sea ice thickness inversion are captured, allowing dynamic adjustment of feature weights to achieve precise inversion of sea ice thickness. Experimental results demonstrate that the SIT-TransNET method significantly improves the accuracy of sea ice thickness inversion, with a coefficient of determination (R²) reaching 0.96 and a root mean square error (RMSE) of 6 cm. This method proves suitable for sea ice thickness inversion and provides an effective technical means for large-scale sea ice thickness monitoring and climate change research.

韩彦岭, 朱晓俊, 王静, 潘海燕, 张云. 基于SIT-TransNET的被动微波海冰厚度反演[J]. 计算机工程, doi: 10.19678/j.issn.1000-3428.0070709.

HAN Yanling , ZHU Xiaojun , WANG Jing , PAN Haiyan, ZHANG Yun. Passive microwave sea ice thickness inversion based on SIT-TransNET[J]. Computer Engineering, doi: 10.19678/j.issn.1000-3428.0070709.

参考文献

[1] YE Y F, WANG X, SUN S Z, et al. A new sea ice concentration retrieval algorithm from thermal infrared imagery [J]. Int J Digit Earth, 2024, 17（1）.
[2] KAPSCH M L, GRAVERSEN R G, TJERNSTRöM M, etal. The Effect of Downwelling Longwave and Shortwave Radiation on Arctic Summer Sea Ice [J]. J Climate, 2016, 29（3）: 1143-59.
[3] 于淼.利用 RadarSat-2 SAR 影像反演北极夏季海冰厚度的研究[D].大连理工大学, 2018. YU Miao, Study on sea ice thickness retrieval in Arctic summer using RadarSat-2 SAR imagery, [D]. Dalian University of Technology, 2018.
[4] Shamshiri R, Eide E, Høyland K V. Spatio-temporal distribution of sea-ice thickness using a machine learning approach with Google Earth Engine and Sentinel-1 GRD data[J]. Remote Sensing of Environment, 2022, 270: 112851.
[5] Zhao Y, Sun Z, Li W, et al. Sea Ice Type and Thickness Identification Based on Vibration Sensor Networks and Machine Learning[J]. IEEE Transactions on Instrumentation and Measurement, 2023, 72: 1-11.
[6] Koenig L, Martin S, Studinger M, et al. Polar airborne observations fill gap in satellite data[J]. Eos, Transactions American Geophysical Union, 2010, 91（38）: 333-334.
[7] Martin S, Drucker R, Kwok R, et al. Estimation of the thin ice thickness and heat flux for the Chukchi Sea Alaskan coast polynya from Special Sensor Microwave/Imager data, 1990–2001[J]. Journal of Geophysical Research: Oceans, 2004, 109（C10）.
[8] Yu Y, Rothrock D A. Thin ice thickness from satellite thermal imagery[J]. Journal of Geophysical Research: Oceans, 1996, 101（C11）: 25753-25766.
[9] 朱星源,苏洁,宋梅,等.基于 MODIS 数据的渤海海冰厚度反演算法优化[J].海洋学报,2022,44（12）:70-83. ZHU Xingyuan, SU Jie, SONG Mei, et al. Optimization of the Bohai Sea ice thickness retrieval algorithm based on MODIS data [J]. Haiyang Xuebao, 2023, 44(12): 70-83. 10.12284/hyxb2022141
[10] 郭井学.基于电磁感应理论的极地海冰厚度探测研究[D]. 吉林大学,2007. GUO Jinxue. Research of electromagnetic-inductive theory on the measurements of polar sea-ice thickness, [D]. Jilin University, 2007.
[11] Karvonen J, Similä M, Hallikainen M, et al. Estimation of Equivalent Deformed Ice Thickness from Baltic Sea Ice[C]. 2005.
[12] Zhang X, Dierking W, Zhang J, et al. Retrieval of the thickness of undeformed sea ice from simulated C-band compact polarimetric SAR images[J]. The Cryosphere, 2016, 10（4）: 1529-1545.
[13] 于淼,卢鹏,李志军,等.基于 SAR 图像纹理的北极海冰厚度的反演研究[J].极地研究,2018,30（03）:329-337. YU Miao, LU Peng, LI Zhijun, et al. Arctic sea ice thickness retrieval based on SAR image texture feature [J]. Chinese Journal of Polar Research, 2018, 30(3): 329. 10.13679/j.jdyj.20170035
[14] Naoki K, Ukita J, Nishio F, et al. Thin sea ice thickness as inferred from passive microwave and in situ observations[J]. Journal of Geophysical Research: Oceans, 2008, 113（C2）.
[15] Herbert C, Munoz-Martin J F, Llaveria D, et al. Sea ice thickness estimation based on regression neural networks using L-band microwave radiometry data from the FSSCat mission[J]. Remote Sensing, 2021, 13（7）: 1366.
[16] Tateyama K, Enomoto H, Toyota T, et al. Sea ice thickness estimated from passive microwave radiometers[J]. Polar meteorology and glaciology, 2002, 16: 15-31.
[17] Junhwa C, Cheol H K. Retrieval of daily sea ice thickness from AMSR2 passive microwave data using ensemble convolutional neural networks[J]. GIScience & Remote Sensing,2021,58（6）:812-830.
[18] 王志勇,张梦悦,于亚冉,等.一种融合纹理特征与 NDVI 的随机森林海冰精细分类方法[J].海洋学报, 2021. WANG Zhiyong, ZHANG Mengyue, YU Yaran, et al. A fine classification method for sea ice based on random forest combining texture feature and NDVI [J]. Haiyang Xuebao, 2021
[19] Khaleghian, S., Ullah, H., Kræmer, T., Hughes, N., Eltoft, T., & Marinoni, A. (2021). Sea ice classification of SAR imagery based on convolution neural networks. Remote Sensing, 13(9), 1734.
[20] 黄骏捷. 基于主动学习与多特征增强的 GBDT 海冰厚度一体化反演[D]. 上海海洋大学.2024. HUANG Junjie. Integrated inversion of GBDT sea ice thickness based on active learning and multi-feature enhancement[D]. Shanghai Ocean University. 2024.
[21] HY. Li, QY. Yan, WM. Huang. Retrieval of sea ice thickness from FY-3E data using Random Forest method[J]. Advances in Space Research, 2024, 74(1): 130-144.
[22] 宋泉臻 , 陈作钧 , 秦品乐 , 等 . 基于超像素引导的 Transformer 低光图像去噪方法[J].计算机工程，2025. SONG Quanzhen., CHEN Zuojun, QIN Pinle, et al. Superpixel Guide for Transformer Low-light Image Denoising Method. Computer Engineering,2025.
[23] Okuyama A, Imaoka K. Intercalibration of advancedmicrowave scanning radiometer-2 （AMSR2） brightness temperature[J]. IEEE Transactions on Geoscience and Remote Sensing, 2015, 53（8）: 4568-4577.
[24] Tian-Kunze X, Kaleschke L, Maaß N, et al. SMOS-derived thin sea ice thickness: algorithm baseline, product specifications and initial verification[J]. The Cryosphere, 2014, 8（3）: 997-1018.
[25] 郭凤莲,赵仁宇,王维滨.无源微波遥感技术在海冰测厚中的应用研究[J].遥感学报,2000,（02）:112-117. GUO Fenglian, ZHAO Renyu, WANG Weibin. Application of passive microwave remote sensing to sea ice thickness measurement [J]. Journal of Remote Sensing, 2000, 4(2): 112-117.
[26] 刘起东,刘超越,邱紫鑫,等. 基于时间感知Transformer的交通流预测方法[J]. 计算机科学,2023:1-13. LIU Qidong, LIU Chaoyue, QIU Zixin, et al. Traffic flow prediction method based on time-aware Transformer[J]. Computer Science, 2023:1-13.
[27] Petkovski E, Marri I, Cristaldi L, et al. State of Health Estimation Procedure for Lithium-Ion Batteries Using Partial Discharge Data and Support Vector Regression[J]. Energies, 2023,17(1): 206.
[28] Bilei G, Yining W, Yanlin S. Fault diagnosis method for hydro-power plants with Bi-LSTM knowledge graph aided by attention scheme[J]. Journal of Vibroengineering,2023,25(8):1629-1641.
[29] Ernesto M P J, Caisse A, Silva A J C. Hyperspectral dimensionality reduction based on SAE-1DCNN feature selection approach[J]. Applied Geomatics,2023,15(4):991-1004.
[30] Gao Y, Xia X, Guo Y. A Thermal Error Prediction Method of High-Speed Motorized Spindle Based on Pelican Optimization Algorithm and CNN-LSTM[J].Applied Sciences,2023,14(1):381.
[31] Comiso J C, Cavalieri D J, Markus T. Sea ice concentration, ice temperature, and snow depth using AMSR-E data[J]. IEEE Transactions on Geoscience and Remote Sensing, 2003, 41（2）: 243-252.

选择文件类型/文献管理软件名称

选择包含的内容