Identification of glaciers using fully polarimetric SAR data based on deep-learning

被引：0
作者：
Fan J. [1 ]
Ke C. [1 ]
Yao G. [1 ]
Wang Z. [1 ]
机构：
[1] School of Geographic and Oceanographic Science, Nanjing University, Nanjing
来源：
National Remote Sensing Bulletin | 2023年 / 27卷 / 09期
基金：
中国国家自然科学基金;
关键词：
ALOS2-PALSAR; deep learning; glaciers; Himalayas; image segmentation; polarimetric decomposition; remote sensing;
D O I：
10.11834/jrs.20221541
中图分类号：
学科分类号：
摘要：
Glacier identification is important for monitoring water resources and climate change in surrounding areas. Although optical images have achieved high accuracy in glacier boundary identification, optical images are affected by cloud cover, and reproducing information under the clouds is difficult. Fully polarized SAR images contain rich features, and deep learning can fully exploit image information. Therefore, using fully polarized SAR images combined with deep learning can compensate for the lack of optical images and obtain accurate glacier recognition results. In this paper, VGG16-unet (VGG16 combined with U-net) is used to identify glaciers based on ALOS2-PALSAR fully polarized images of the western part of the Himalayas. The features include the diagonal elements of the polarization coherence matrix, Freeman-Durden, H/A/α, Pauli, VanZyl, and Yamaguchi polarization decomposition parameters totaling 19 features. To make full use of the image information, these features are analyzed and combined, and the glacier recognition accuracies are compared to select the best features. Given evident differences between glacier and nonglacier topography, elevation, slope, and local incidence angle are combined with polarization features as auxiliary features. Comparing the classification accuracy of different polarization features reveals the accuracy of Pauli, Freeman-Durden, VanZyl, and Yamaguchi features based on physical characteristics is higher, among which Pauli features have the highest accuracy with an Overall Accuracy (OA) of 92.54% and an average user intersection ratio (mIoU) of 78.78%. The OA is improved to 94.34%, and the mIoU is improved to 82.35% after adding the topographic data. In order to improve the recognition accuracy of glaciers further, a feature cross-combination approach is proposed, and results show the OA of the combination reaches 94.98%, and the mIoU reaches 85.67%, which are 0.64% and 3.32% higher than the classification accuracy of Pauli features, respectively. Selecting the best feature combination method and combining with deep learning plays an important role in improving the accuracy of glacier recognition, and the use of neural networks combined with fully polarized SAR images can effectively compensate for the shortcomings of optical images in glacier boundary identification. © 2023 Science Press. All rights reserved.
引用
收藏
页码：2098 / 2113
页数：15
相关论文
共 28 条
[11]  
Nie Y, Zhang Y L, Liu L S, Zhang J P., Monitoring glacier change based on remote sensing in the Mt. Qomolangma national nature preserve, 1976-2006, Acta Geographica Sinica, 65, 1, pp. 13-28, (2010)
[12]  
Parrella G, Hajnsek I, Papathanassiou K P., Polarimetric decomposition of L-band PolSAR backscattering over the austfonna ice cap, IEEE Transactions on Geoscience and Remote Sensing, 54, 3, pp. 1267-1281, (2016)
[13]  
Paul F, Bolch T, Briggs K, Kaab A, McMillan M, McNabb R, Nagler T, Nuth C, Rastner P, Strozzi T, Wuite J., Error sources and guidelines for quality assessment of glacier area, elevation change, and velocity products derived from satellite data in the Glaciers_cci project, Remote Sensing of Environment, 203, pp. 256-275, (2017)
[14]  
Paul F, Bolch T, Kaab A, Nagler T, Nuth C, Scharrer K, Shepherd A, Strozzi T, Ticconi F, Bhambri R, Berthier E, Bevan S, Gourmelen N, Heid T, Jeong S, Kunz M, Lauknes T R, Luckman A, Boncori J P M, Moholdt G, Muir A, Neelmeijer J, Rankl M, Van Looy J, Van Niel T., The glaciers climate change initiative: methods for creating glacier area, elevation change and velocity products, Remote Sensing of Environment, 162, pp. 408-426, (2015)
[15]  
Ronneberger O, Fischer P, Brox T., U-net: convolutional networks for biomedical image segmentation, 18th International Conference on Medical Image Computing and Computer-Assisted Intervention, pp. 234-241, (2015)
[16]  
Scherler D, Bookhagen B, Strecker M R., Spatially variable response of Himalayan glaciers to climate change affected by debris cover, Nature Geoscience, 4, 3, pp. 156-159, (2011)
[17]  
Sharma J J, Hajnsek I, Papathanassiou K P, Moreira A., Polarimetric decomposition over glacier ice using long-wavelength airborne polSAR, IEEE Transactions on Geoscience and Remote Sensing, 49, 1, pp. 519-535, (2011)
[18]  
Shi J C, Dozier J., Measurements of snow-and glacier-covered areas with single-polarization SAR, Annals of Glaciology, 17, pp. 72-76, (1993)
[19]  
Simonyan K, Zisserman A., Very deep convolutional networks for large-scale image recognition, (2015)
[20]  
Singh G, Venkataraman G, Yamaguchi Y, Park S E., Capability assessment of fully polarimetric ALOS-PALSAR data for discriminating wet snow from other scattering types in mountainous regions, IEEE Transactions on Geoscience and Remote Sensing, 52, 2, pp. 1177-1196, (2014)
123