Large-Scale Rice Mapping Using Multi-Task Spatiotemporal Deep Learning and Sentinel-1 SAR Time Series

被引:43
作者
Lin, Zhixian [1 ]
Zhong, Renhai [1 ,2 ]
Xiong, Xingguo [1 ]
Guo, Changqiang [1 ]
Xu, Jinfan [1 ]
Zhu, Yue [1 ]
Xu, Jialu [1 ]
Ying, Yibin [1 ]
Ting, K. C. [2 ,3 ]
Huang, Jingfeng [4 ]
Lin, Tao [1 ]
机构
[1] Zhejiang Univ, Coll Biosyst Engn & Food Sci, Hangzhou 310058, Peoples R China
[2] Zhejiang Univ, Int Campus, Haining 314400, Peoples R China
[3] Univ Illinois, Dept Agr & Biol Engn, Urbana, IL 61801 USA
[4] Zhejiang Univ, Inst Appl Remote Sensing & Informat Technol, Hangzhou 310058, Peoples R China
基金
中国国家自然科学基金;
关键词
spatiotemporal analysis; deep learning; long short-term memory; multi-task learning; rice mapping; Sentinel-1; LAND-COVER; CROP; CLASSIFICATION; DELINEATION; SYSTEM; US;
D O I
10.3390/rs14030699
中图分类号
X [环境科学、安全科学];
学科分类号
08 ; 0830 ;
摘要
Timely and accurate cropland information at large spatial scales can improve crop management and support the government in decision making. Mapping the spatial extent and distribution of crops on a large spatial scale is challenging work due to the spatial variability. A multi-task spatiotemporal deep learning model, named LSTM-MTL, was developed in this study for large-scale rice mapping by utilizing time-series Sentinel-1 SAR data. The model showed a reasonable rice classification accuracy in the major rice production areas of the U.S. (OA = 98.3%, F1 score = 0.804), even when it only utilized SAR data. The model learned region-specific and common features simultaneously, and yielded a significant improved performance compared with RF and AtBiLSTM in both global and local training scenarios. We found that the LSTM-MTL model achieved a regional F1 score up to 10% higher than both global and local baseline models. The results demonstrated that the consideration of spatial variability via LSTM-MTL approach yielded an improved crop classification performance at a large spatial scale. We analyzed the input-output relationship through gradient backpropagation and found that low VH value in the early period and high VH value in the latter period were critical for rice classification. The results of in-season analysis showed that the model was able to yield a high accuracy (F1 score = 0.746) two months before rice maturity. The integration between multi-task learning and multi-temporal deep learning approach provides a promising approach for crop mapping at large spatial scales.
引用
收藏
页数:21
相关论文
共 64 条
[1]  
[Anonymous], 2020, USDA NASS QUICK STAT
[2]  
[Anonymous], 2021, IEEE Trans. Broadcast.
[3]  
[Anonymous], 2018, The State of Food Security and Nutrition in the World: Building Climate Resilience for Food Security and Nutrition
[4]   SAR data for tropical forest disturbance alerts in French Guiana: Benefit over optical imagery [J].
Ballere, Marie ;
Bouvet, Alexandre ;
Mermoz, Stephane ;
Le Toan, Thuy ;
Koleck, Thierry ;
Bedeau, Caroline ;
Andre, Mathilde ;
Forestier, Elodie ;
Frison, Pierre-Louis ;
Lardeux, Cedric .
REMOTE SENSING OF ENVIRONMENT, 2021, 252
[5]   Monitoring US agriculture: the US Department of Agriculture, National Agricultural Statistics Service, Cropland Data Layer Program [J].
Boryan, Claire ;
Yang, Zhengwei ;
Mueller, Rick ;
Craig, Mike .
GEOCARTO INTERNATIONAL, 2011, 26 (05) :341-358
[6]  
BOUMAN BAM, 1995, NETH J AGR SCI, V43, P143
[7]  
Breiman L, 1996, MACH LEARN, V24, P123, DOI 10.1023/A:1018054314350
[8]   A high-performance and in-season classification system of field-level crop types using time-series Landsat data and a machine learning approach [J].
Cai, Yaping ;
Guan, Kaiyu ;
Peng, Jian ;
Wang, Shaowen ;
Seifert, Christopher ;
Wardlow, Brian ;
Li, Zhan .
REMOTE SENSING OF ENVIRONMENT, 2018, 210 :35-47
[9]   Early-season mapping of winter wheat in China based on Landsat and Sentinel images [J].
Dong, Jie ;
Fu, Yangyang ;
Wang, Jingjing ;
Tian, Haifeng ;
Fu, Shan ;
Niu, Zheng ;
Han, Wei ;
Zheng, Yi ;
Huang, Jianxi ;
Yuan, Wenping .
EARTH SYSTEM SCIENCE DATA, 2020, 12 (04) :3081-3095
[10]   Mapping paddy rice planting area in northeastern Asia with Landsat 8 images, phenology-based algorithm and Google Earth Engine [J].
Dong, Jinwei ;
Xiao, Xiangming ;
Menarguez, Michael A. ;
Zhang, Geli ;
Qin, Yuanwei ;
Thau, David ;
Biradar, Chandrashekhar ;
Moore, Berrien, III .
REMOTE SENSING OF ENVIRONMENT, 2016, 185 :142-154