利用深度学习提取青藏高原湖泊季节性面积变化的理论和方法研究

青藏高原作为亚洲水塔，水储量十分丰富，同时由于青藏高原的高海拔环境特性与较少的人类活动，使得青藏高原湖泊可以作为响应气候变化的指示器。迄今为止，大多数研究利用光学遥感数据揭示了自20世纪70年代以来青藏高原湖泊面积的年际变化。然而受年内不同季节时期的降雨模式影响，青藏高原湖泊面积也会发生季节性变化。因此，探究湖泊面积季节性变化对揭示青藏高原地区湖泊面积变化的驱动因素具有一定研究意义。但是由于每年高质量的光学影像数据量较少，监测青藏高原湖泊面积季节性变化仍存在困难。

目前光学影像因容易受到云层污染而常用于研究高原湖泊年际变化，对于湖泊面积的季节性变化，若仅利用雷达影像进行研究，得到的时序结果与水位数据相关性较差。针对上述问题，本文基于雷达影像和光学影像各自的优势，提出一种利用多源融合影像进行湖泊识别的数据处理策略，从而得到高时间分辨率的湖泊面积变化。

与传统湖泊提取方法相比，深度学习算法具有高效准确的优势。通过对包括DeepLabv3+、UNet、ResUNet、DeepUNet、SegNet等主流深度学习模型的评估，发现CloudNet+模型在影像分割方面表现最为优异，在F1_Score（0.985）、Precision（0.981）、Recall（0.990）和mIoU（0.938）指标中精度最高。为了使该模型能更充分地利用多源融合影像中的光谱与空间信息，本文在该模型基础上提出了一种引入注意力机制的改进模型——AttsCloudNet+，可从多源融合影像中提取夏秋两季的湖泊。此外，本文还利用自主研发的轻量级LaeNet模型对冬春两季光学影像中的湖泊进行提取，从而获取完整的湖泊季节性面积变化时序。AttsCloudNet+与LaeNet在处理各自适用类型数据源的过程中与其他主流深度学习卷积神经网络模型相比有更好的性能，其中AttsCloudNet+在各项评价指标包括F1_Score（0.986）、Precision（0.982）、Recall（0.992）和mIoU（0.945）方面表现更佳。

为了验证湖泊提取结果的准确性，本文使用湖岸线实测GPS数据，来评价AttsCloudNet+和LaeNet模型的预测结果。其中，AttsCloudNet+模型预测的融合影像中的湖泊边界，其平均RMSE和MAE的平均值分别为21.63 m和16.59 m。LaeNet模型对光学影像中湖泊边界的预测精度，平均RMSE和MAE分别为24.88 m和19.11 m。与此同时，本文还将DeepLabv3+和UNet的预测结果同实测数据进行了对比，其中DeepLabv3+和UNet对融合影像提取结果的平均RMSE为99.53 m和91.10 m，平均MAE为75.99 m和64.87 m；对于光学影像，DeepLabv3+和UNet的平均RMSE为40.34 m和27.27 m，平均MAE为29.77 m和21.65 m。因此，本文的AttsCloudNet+和LaeNet模型整体精度优于常用主流深度学习模型。此外，本文在对融合影像进行预测时，通过补充不同获取时间的少量高质量光学影像，发现AttsCloudNet+同DeepLabv3+、UNet模型相比更具稳健性，其平均精度为23.77 m。

本文还将所得青藏高原湖泊的季节性面积变化时间序列与LEGOS湖泊水位数据集进行相关性分析，其相关系数R²为0.80，与前人结果相比提高约2.3倍。本文对所得15个湖泊季节性面积变化时间序列与其他公开水位和面积数据集进行对比分析，发现青海湖等9个湖泊在2017年和2018年出现了连续增长；同时发现季节性湖泊面积变化还表现出一定的空间差异，位于青藏高原中部与南部的湖泊呈现出夏季扩张和冬季收缩的趋势；而小部分湖泊如青藏高原北部的阿其克库勒湖则没有发现明显的季节变化趋势。

[1] IMMERZEEL W W, VAN BEEK L P, BIERKENS M F. Climate change will affect the Asian water towers [J]. science, 2010, 328(5984): 1382-1385.
[2] KUANG X, JIAO J J. Review on climate change on the Tibetan Plateau during the last half century [J]. Journal of Geophysical Research: Atmospheres, 2016, 121(8): 3979-4007.
[3] MA Y, MA W, ZHONG L, et al. Monitoring and modeling the Tibetan Plateau’s climate system and its impact on East Asia [J]. Scientific Reports, 2017, 7(1): 1-6.
[4] SONG C, HUANG B, KE L. Modeling and analysis of lake water storage changes on the Tibetan Plateau using multi-mission satellite data [J]. Remote Sensing of Environment, 2013, 135: 25-35.
[5] ZHANG G, YAO T, CHEN W, et al. Regional differences of lake evolution across China during 1960s–2015 and its natural and anthropogenic causes [J]. Remote Sensing of Environment, 2019, 221: 386-404.
[6] ZHANG G, YAO T, PIAO S, et al. Extensive and drastically different alpine lake changes on Asia's high plateaus during the past four decades [J]. Geophysical Research Letters, 2017, 44(1): 252-260.
[7] KUANG X, JIAO J J. Review on climate change on the Tibetan plateau during the last half century [J]. Journal of Geophysical Research, 2016, 121(8): 3979-4007.
[8] LIU J, WANG S, YU S, et al. Climate warming and growth of high-elevation inland lakes on the Tibetan Plateau [J]. Global and Planetary Change, 2009, 67(3-4): 209-217.
[9] LIU X, CHEN B. Climatic warming in the Tibetan Plateau during recent decades [J]. International Journal of Climatology: A Journal of the Royal Meteorological Society, 2000, 20(14): 1729-1742.
[10] MAUSSION F, SCHERER D, MöLG T, et al. Precipitation seasonality and variability over the Tibetan Plateau as resolved by the high asia reanalysis [J]. Journal of Climate, 2014, 27(5): 1910-1927.
[11] ZHANG G, YAO T, XIE H, et al. Response of Tibetan Plateau lakes to climate change: trends, patterns, and mechanisms [J]. Earth-Science Reviews, 2020, 208.
[12] 张丁玲. 青藏高原水资源时空变化特征的研究 [D]; 兰州大学, 2013.
[13] 李生辰, 徐亮, 郭英香, 等. 近34a青藏高原年降水变化及其分区 [J]. 中国沙漠, 2007, (02): 307-314.
[14] 王可丽, 江灏, 赵红岩. 西风带与季风对中国西北地区的水汽输送 [J]. 水科学进展, 2005, (03): 432-438.
[15] 马荣华, 杨桂山, 段洪涛, 等. 中国湖泊的数量、面积与空间分布 [J]. 中国科学:地球科学, 2011, 41(03): 394-401.
[16] 朱立平, 张国庆, 杨瑞敏, 等. 青藏高原最近40年湖泊变化的主要表现与发展趋势 [J]. 中国科学院院刊, 2019, 34(11): 1254-1263.
[17] 闫立娟, 郑绵平, 魏乐军. 近40年来青藏高原湖泊变迁及其对气候变化的响应 [J]. 地学前缘, 2016, 23(04): 310-323.
[18] PEKEL J F, COTTAM A, GORELICK N, et al. High-resolution mapping of global surface water and its long-term changes [J]. Nature, 2016, 540(7633): 418-422.
[19] 万玮, 肖鹏峰, 冯学智, 等. 卫星遥感监测近30年来青藏高原湖泊变化 [J]. 科学通报, 2014, 59(08): 701-714.
[20] 姜永见, 李世杰, 沈德福, 等. 青藏高原近40年来气候变化特征及湖泊环境响应 [J]. 地理科学, 2012, 32(12): 1503-1512.
[21] 李均力, 盛永伟, 骆剑承, 等. 青藏高原内陆湖泊变化的遥感制图 [J]. 湖泊科学, 2011, 23(03): 311-320.
[22] YU J, ZHANG G, YAO T, et al. Developing daily cloud-free snow composite products from MODIS Terra–Aqua and IMS for the Tibetan Plateau [J]. IEEE Transactions on Geoscience and Remote Sensing, 2016, 54(4): 2171-2180.
[23] ZHANG Y, ZHANG G, ZHU T. Seasonal cycles of lakes on the Tibetan Plateau detected by Sentinel-1 SAR data [J]. Sci Total Environ, 2020, 703: 135563.
[24] YAO F, WANG J, WANG C, et al. Constructing long-term high-frequency time series of global lake and reservoir areas using Landsat imagery [J]. Remote Sensing of Environment, 2019, 232.
[25] DAI K, WEN N, FAN X, et al. Seasonal changes of glacier lakes in Tibetan Plateau revealed by multipolarization SAR Data [J]. IEEE Geoscience and Remote Sensing Letters, 2022, 19: 1-5.
[26] 毕海芸, 王思远, 曾江源, 等. 基于TM影像的几种常用水体提取方法的比较和分析 [J]. 遥感信息, 2012, 27(05): 77-82.
[27] 王海波, 马明国. 基于遥感的湖泊水域动态变化监测研究进展 [J]. 遥感技术与应用, 2009, 24(05): 674-684.
[28] MCFEETERS S K. The use of the Normalized Difference Water Index (NDWI) in the delineation of open water features [J]. International Journal of Remote Sensing, 2007, 17(7): 1425-1432.
[29] 徐涵秋. 利用改进的归一化差异水体指数(MNDWI)提取水体信息的研究 [J]. 遥感学报, 2005, (05): 589-595.
[30] FEYISA G L, MEILBY H, FENSHOLT R, et al. Automated Water Extraction Index: A new technique for surface water mapping using Landsat imagery [J]. Remote Sensing of Environment, 2014, 140: 23-35.
[31] ABBAS A, MINALLH N, AHMAD N, et al. K-Means and ISODATA clustering algorithms for landcover classification using remote sensing [J]. Sindh University Research Journal, 2016, 48.
[32] ALTUNKAYNAK A. Forecasting surface water level fluctuations of lake van by Artificial Neural Networks [J]. Water Resources Management, 2006, 21(2): 399-408.
[33] CORTES C, VAPNIK V. Support-Vector Networks [J]. Machine Learning, 1995, 20(3): 273-297.
[34] VINCENZI S, ZUCCHETTA M, FRANZOI P, et al. Application of a Random Forest algorithm to predict spatial distribution of the potential yield of ruditapes philippinarum in the venice lagoon, Italy [J]. Ecological Modelling, 2011, 222(8): 1471-1478.
[35] MELGANI F, BRUZZONE L. Classification of hyperspectral remote sensing images with support vector machines [J]. IEEE Transactions on Geoscience and Remote Sensing, 2004, 42(8): 1778-1790.
[36] PAL M. Random forest classifier for remote sensing classification [J]. International Journal of Remote Sensing, 2005, 26(1): 217-222.
[37] BELGIU M, DRĂGU L. Random forest in remote sensing: A review of applications and future directions [J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2016, 114: 24-31.
[38] BREIMAN L. Random forests [J]. Machine Learning, 2001, 45(1): 5-32.
[39] MOUNTRAKIS G, IM J, OGOLE C. Support vector machines in remote sensing: A review [J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2011, 66(3): 247-259.
[40] 高子恒, 丁炜, 何静. 基于随机森林算法和MODIS数据的日喀则地区土地覆盖分类与动态监测 [J]. 安徽农业科学, 2020, 48(16): 1-12.
[41] 王琼, 王君波, 郭俊钰, 等. 基于支持向量机的色林错湖冰提取及时空分布 [J]. 载人航天, 2019, 25(06): 789-798.
[42] QAYYUM N, GHUFFAR S, AHMAD H M, et al. Glacial lakes mapping using multi satellite Planetscope imagery and deep learning [J]. ISPRS International Journal of Geo-Information, 2020, 9(10).
[43] 金永涛, 杨秀峰, 高涛, 等. 基于面向对象与深度学习的典型地物提取 [J]. 国土资源遥感, 2018, 30(01): 22-29.
[44] 杜敬. 基于深度学习的无人机遥感影像水体识别 [J]. 江西科学, 2017, 35(01): 158-161+170.
[45] 党宇, 张继贤, 邓喀中, 等. 基于深度学习AlexNet的遥感影像地表覆盖分类评价研究 [J]. 地球信息科学学报, 2017, 19(11): 1530-1537.
[46] 周岩, 董金玮. 陆表水体遥感监测研究进展 [J]. 地球信息科学学报, 2019, 21(11): 1768-1778.
[47] 陈斌, 王宏志, 徐新良, 等. 深度学习GoogleNet模型支持下的中分辨率遥感影像自动分类 [J]. 测绘通报, 2019, (06): 29-33.
[48] 苏龙飞, 李振轩, 高飞, 等. 遥感影像水体提取研究综述 [J]. 国土资源遥感, 2021, 33(01): 9-19.
[49] 王鑫, 徐明君, 李可, 等. 一种有效的高分辨率遥感影像水体提取方法 [J]. 计算机工程与应用, 2019, 55(20): 145-151.
[50] WENG L, XU Y, XIA M, et al. Water areas segmentation from remote sensing images using a separable residual SegNet network [J]. ISPRS International Journal of Geo-Information, 2020, 9(4).
[51] 何海清, 杜敬, 陈婷, 等. 结合水体指数与卷积神经网络的遥感水体提取 [J]. 遥感信息, 2017, 32(05): 82-86.
[52] 陈前, 郑利娟, 李小娟, 等. 基于深度学习的高分遥感影像水体提取模型研究 [J]. 地理与地理信息科学, 2019, 35(04): 43-49.
[53] KANG S, XU Y, YOU Q, et al. Review of climate and cryospheric change in the Tibetan Plateau [J]. Environmental Research Letters, 2010, 5(1).
[54] 韦志刚, 黄荣辉, 董文杰. 青藏高原气温和降水的年际和年代际变化 [J]. 大气科学, 2003, (02): 157-170.
[55] WANG W, XIANG Y, GAO Y, et al. Rapid expansion of glacial lakes caused by climate and glacier retreat in the Central Himalayas [J]. Hydrological Processes, 2015, 29(6): 859-874.
[56] ZHANG G, LUO W, CHEN W, et al. A robust but variable lake expansion on the Tibetan Plateau [J]. Science Bulletin, 2019, 64(18): 1306-1309.
[57] 杨珂含. 基于多源多时相卫星影像的青藏高原湖泊面积动态监测 [D]; 中国科学院大学(中国科学院遥感与数字地球研究所), 2017.
[58] 孙芳蒂. 中国主要湖泊面积2000-2010年动态遥感监测 [D]; 南京大学, 2013.
[59] GORELICK N, HANCHER M, DIXON M, et al. Google Earth Engine: Planetary-scale geospatial analysis for everyone [J]. Remote Sensing of Environment, 2017, 202: 18-27.
[60] MARKHAM B L, STOREY J C, WILLIAMS D L, et al. Landsat sensor performance: history and current status [J]. IEEE transactions on geoscience and remote sensing, 2004, 42(12): 2691-2694.
[61] MARKHAM B L, HELDER D L. Forty-year calibrated record of earth-reflected radiance from Landsat: A review [J]. Remote Sensing of Environment, 2012, 122: 30-40.
[62] CRéTAUX J F, ARSEN A, CALMANT S, et al. SOLS: A lake database to monitor in the near real time water level and storage variations from remote sensing data [J]. Advances in Space Research, 2011, 47(9): 1497-1507.
[63] LI X, LONG D, HUANG Q, et al. High-temporal-resolution water level and storage change data sets for lakes on the Tibetan Plateau during 2000–2017 using multiple altimetric missions and Landsat-derived lake shoreline positions [J]. Earth System Science Data, 2019, 11(4): 1603-1627.
[64] KRIZHEVSKY A, SUTSKEVER I, HINTON G E. ImageNet classification with deep convolutional neural networks; proceedings of the Advances in Neural Information Processing Systems, F, 2012 [C].
[65] LECUN Y, BENGIO Y, HINTON G. Deep learning [J]. Nature, 2015, 521(7553): 436-444.
[66] WAGNER F H, SANCHEZ A, TARABALKA Y, et al. Using the U-net convolutional network to map forest types and disturbance in the Atlantic rainforest with very high resolution images [J]. Remote Sensing in Ecology and Conservation, 2019, 5(4): 360-375.
[67] ZHOU Z, RAHMAN SIDDIQUEE M M, TAJBAKHSH N, et al. Unet++: A nested u-net architecture for medical image segmentation [Z]. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). 2018: 3-11.
[68] DIAKOGIANNIS F I, WALDNER F, CACCETTA P, et al. ResUNet-a: A deep learning framework for semantic segmentation of remotely sensed data [J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2020, 162: 94-114.
[69] LI R, LIU W, YANG L, et al. DeepUNet: A deep fully convolutional network for pixel-level sea-land segmentation [J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2018, 11(11): 3954-3962.
[70] CHEN L C, ZHU Y, PAPANDREOU G, et al. Encoder-decoder with atrous separable convolution for semantic image segmentation [Z]. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). 2018: 833-851.10.1007/978-3-030-01234-2_49
[71] SIMONYAN K, ZISSERMAN A. Very deep convolutional networks for large-scale image recognition [J]. CoRR, 2015, abs/1409.1556.
[72] BADRINARAYANAN V, KENDALL A, CIPOLLA R. SegNet: A deep convolutional encoder-decoder architecture for image segmentation [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(12): 2481-2495.
[73] WOO S, PARK J, LEE J Y, et al. CBAM: Convolutional block attention module [Z]. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). 2018: 3-19
[74] MOHAJERANI S, SAEEDI P. Cloud-Net+: A cloud segmentation CNN for Landsat 8 remote sensing imagery optimized with Filtered Jaccard Loss function [J]. ArXiv, 2020, abs/2001.08768.
[75] LIU W, CHEN X, RAN J, et al. LaeNet: A novel lightweight multitask CNN for automatically extracting lake area and shoreline from remote sensing images [J]. Remote Sensing, 2020, 13(1).