融合自动样本生成与可解释学习的风云三号时序NDVI冬小麦识别*

doi:10.7523/j.ucas.2026.010

中国科学院大学学报

• •

融合自动样本生成与可解释学习的风云三号时序NDVI冬小麦识别^*

王军强¹, 孙震辉^2,†, 孟庆岩^3,4,5, 张琳琳^3,4,5, 孙云晓²

1.生态环境部土壤与农业农村生态环境监管技术中心, 北京 100012;
2.天津城建大学地质与测绘学院, 天津 300384;
3.中国科学院空天信息创新研究院, 北京 100101;
4.中国科学院大学, 北京 100049;
5.海南空天信息研究院海南省地球观测重点实验室, 海南三亚 572029

收稿日期:2025-11-21 修回日期:2026-03-23 发布日期:2026-04-21
通讯作者: ^†E-mail：sunzh@tcu.edu.cn
基金资助:
^*生态环境部土壤与农业农村生态环境监管技术中心双碳课题(2024-11),风云三号03批气象卫星工程地面应用系统生态监测评估应用项目（第一期）(ZQC-R22227)和天津市科学技术普及项目（24KPHDRC00240）资助

Winter wheat identification from FengYun-3 NDVI time series with automatic sample generation and interpretable learning

WANG Junqiang¹, SUN Zhenhui², MENG Qingyan^3,4,5, ZHANG Linlin^3,4,5, SUN Yunxiao²

1 Technical Centre for Soil, Agriculture and Rural Ecology and Environment, Ministry of Ecology and Environment, Beijing, 100012, China;
2 School of Geology and Geomatics, Tianjin Chengjian University, Tianjin 300384, China;
3 Aerospace Information Research Institute, Chinese Academy of Sciences, Beijing 100101, China;
4 University of Chinese Academy of Sciences, Beijing 100049, China;
5 Key Laboratory of Earth Observation of Hainan Province, Hainan Aerospace Information Research Institute, Sanya 572029, Hainan, China

Received:2025-11-21 Revised:2026-03-23 Published:2026-04-21

摘要/Abstract

摘要： 及时、准确获取冬小麦信息至关重要,高质量样本的自动生成和识别模型的可解释性依然是制约当前作物提取和制图的重要问题。基于风云三号D星NDVI时序产品,本文提出了一种自动样本生成与可解释CatBoost模型的冬小麦识别方法,高质量分割一切模型(HQ-SAM)用于自动生成高质量样本,Shapley加性解释模型（SHAP）对CatBoost进行可视化解释。结果表明：（1）HQ-SAM可以生成高质量训练样本,降低样本制作成本;（2）鲸鱼优化算法优化后的CatBoost准确度和F1值分别为96.43%和96.36%;（3）SHAP模型分析发现时相的重要性排序及正负贡献方向与冬小麦生长规律一致;（4）基于SHAP模型时相重要性排序结果,前17个时相形成的组合取得了最高精度。本研究为风云气象卫星在农作物提取的应用中提供了技术支撑。

关键词: 风云三号, 样本生成, 模型可解释, 冬小麦, 作物制图

Abstract: Winter wheat is one of the most important staple crops in China and worldwide, and the timely and accurate acquisition of winter wheat information is crucial for agricultural monitoring, yield assessment, and food security. However, the automatic generation of high-quality training samples and the interpretability of classification models remain key challenges limiting the reliability and applicability of current crop identification and mapping approaches. Based on the NDVI time-series product derived from the FengYun-3D meteorological satellite, this study proposes a winter wheat identification method integrating automatic sample generation with an interpretable CatBoost model. A High-Quality Segment Anything Model (HQ-SAM) is introduced to automatically generate high-quality training samples, and effectively reduce labor cost and uncertainty associated with manual sample collection. To further enhance classification performance, the Whale Optimization Algorithm (WOA) is employed to optimize the CatBoost classifier. In addition, the Shapley Additive Explanations (SHAP) method is applied to provide quantitative and visual interpretations of model predictions, enabling analysis of temporal feature contributions. Experimental results demonstrate that: (1) HQ-SAM can reliably generate high-quality training samples for winter wheat identification; (2) the WOA-optimized CatBoost model achieves an overall accuracy of 96.43% and an F1 value of 96.36%; (3) SHAP-based analysis reveals that the importance ranking and contribution directions of temporal phases are consistent with winter wheat phenology; and (4) the combination of the top 17 most important temporal phases yields the highest classification accuracy.

Key words: FengYun-3D, sample generation, model interpretability, winter wheat, crop mapping

中图分类号:

TP79

王军强, 孙震辉, 孟庆岩, 张琳琳, 孙云晓. 融合自动样本生成与可解释学习的风云三号时序NDVI冬小麦识别^*[J]. 中国科学院大学学报, DOI: 10.7523/j.ucas.2026.010.

WANG Junqiang, SUN Zhenhui, MENG Qingyan, ZHANG Linlin, SUN Yunxiao. Winter wheat identification from FengYun-3 NDVI time series with automatic sample generation and interpretable learning[J]. Journal of University of Chinese Academy of Sciences, DOI: 10.7523/j.ucas.2026.010.

参考文献

[1] Chen Y, Zhang Z, Tao F L.Improving regional winter wheat yield estimation through assimilation of phenology and leaf area index from remote sensing data[J]. European Journal of Agronomy, 2018, 101: 163-173. DOI: 10.1016/j.eja.2018.09.006.
[2] Atzberger C.Advances in remote sensing of agriculture: Context description, existing operational monitoring systems and major information needs[J]. Remote Sensing, 2013, 5(2): 949-981. DOI: 10.3390/rs5020949.
[3] 樊香所, 许文波, 范锦龙. FY-3250m分辨率数据的华北平原冬小麦提取[J]. 遥感学报, 2015, 19(4): 586-593. DOI: 10.11834/jrs.20154135.
[4] 冯权泷,任燕,姚晓闯,等.基于多源光学雷达数据融合的黄淮海平原冬小麦识别[J].农业机械学报,2023,54(2):160-168. DOI: 10.6041/j.issn.1000-1298.2023.02.015.
[5] Skakun S, Franch B, Vermote E, et al.Early season large-area winter crop mapping using MODIS NDVI data, growing degree days information and a Gaussian mixture model[J]. Remote Sensing of Environment, 2017, 195: 244-258. DOI: 10.1016/j.rse.2017.04.026.
[6] 田欣媛,张永红,刘睿,等.考虑植被红边信息的多时相Sentinel-2大范围冬小麦提取研究[J].遥感学报,2022,26(10):1988-2000. DOI: 10.11834/jrs.20211359.
[7] Tao J B, Wu W B, Zhou Y, et al.Mapping winter wheat using phenological feature of peak before winter on the North China Plain based on time-series MODIS data[J]. Journal of Integrative Agriculture, 2017, 16(2): 348-359. DOI: 10.1016/S2095-3119(15)61304-1.
[8] Yang B, Wang J L, Li S L, et al.Identifying the spatio-temporal change in winter wheat-summer maize planting structure in the North China Plain between 2001 and 2020[J]. Agronomy, 2023, 13(11): 2712. DOI: 10.3390/agronomy13112712.
[9] 张莎,张佳华,白雲,等.基于MODIS-EVI及物候差异免阈值提取黄淮海平原冬小麦面积[J].农业工程学报,2018,34(11):150-158. DOI: 10.11975/j.issn.1002-6819.2018.11.019.
[10] Wang Y Y, Li G C.Global land cover mapping using annual clear-sky composites from FY3D/MERSI-II[J]. International Journal of Remote Sensing, 2022, 43(2): 510-531. DOI: 10.1080/01431161.2021.2019848.
[11] 钱永兰, 赵晓凤, 张晔萍, 等. FY-3B/D星NDVI与MODIS NDVI对作物长势监测对比[J]. 气象学报, 2023, 81(4):660-672. DOI: 10.11676/qxxb2023.20230012.
[12] 王圆圆, 李贵才. 全球风云三号D星MERSI/NDVI产品质量评估[J]. 气象学报, 2022, 80(1): 124-135. DOI: 10.11676/qxxb2022.007.
[13] Zhang F M, Wang X F, Gao G, et al.Winter wheat identification in China through FY-3D NDVI&EVI satellite data and DNN Fusion[J]. International Journal of Remote Sensing, 2025, 46(1): 300-319. DOI: 10.1080/01431161.2024.2413026.
[14] Gao Z T, Guo D L, Ryu D, et al.Training sample selection for robust multi-year within-season crop classification using machine learning[J]. Computers and Electronics in Agriculture, 2023, 210: 107927. DOI: 10.1016/j.compag.2023.107927.
[15] Son N T, Chen C F, Chen C R, et al.Assessment of Sentinel-1A data for rice crop classification using random forests and support vector machines[J]. Geocarto International, 2018, 33(6): 587-601. DOI: 10.1080/10106049.2017.1289555.
[16] Wang X, Zhang J H, Xun L, et al.Evaluating the effectiveness of machine learning and deep learning models combined time-series satellite data for multiple crop types classification over a large-scale region[J]. Remote Sensing, 2022, 14(10): 2341. DOI: 10.3390/rs14102341.
[17] Zhong L H, Hu L N, Zhou H.Deep learning based multi-temporal crop classification[J]. Remote Sensing of Environment, 2019, 221: 430-443. DOI: 10.1016/j.rse.2018.11.032.
[18] Rodriguez-Galiano V F, Chica-Olmo M, Abarca-Hernandez F, et al. Random Forest classification of Mediterranean land cover using multi-seasonal imagery and multi-seasonal texture[J]. Remote Sensing of Environment, 2012, 121: 93-107. DOI: 10.1016/j.rse.2011.12.003.
[19] Prokhorenkova L, Gusev G, Vorobev A, et al. CatBoost: Unbiased boosting with categorical features[EB/OL].2017: arXiv: 1706.09516. https://arxiv.org/abs/1706.09516.
[20] Hancock J T, Khoshgoftaar T M.CatBoost for big data: An interdisciplinary review[J]. Journal of Big Data, 2020, 7(1): 94. DOI: 10.1186/s40537-020-00369-8.
[21] 黄发明,陈彬,毛达雄,等.基于自筛选深度学习的滑坡易发性预测建模及其可解释性[J].地球科学,2023,48(5):1696-1710. DOI: 10.3799/dqkx.2022.247.
[22] Sun J L, Yan S, Alexandridis T, et al.Enhancing crop mapping through automated sample generation based on segment anything model with medium-resolution satellite imagery[J]. Remote Sensing, 2024, 16(9): 1505. DOI: 10.3390/rs16091505.
[23] 全国卫星气象与空间天气标准化技术委员会. 气象卫星数据文件名命名规范: QX/T 387-2017[S]. 北京:气象出版社, 2018:9-11.
[24] Wardlow B D, Egbert S L, Kastens J H.Analysis of time-series MODIS 250 m vegetation index data for crop classification in the U.S. Central Great Plains[J]. Remote Sensing of Environment, 2007, 108(3): 290-310. DOI: 10.1016/j.rse.2006.11.021.
[25] 李方杰,任建强,吴尚蓉,等.NDVI时序相似性对冬小麦种植面积总量控制的制图精度影响[J].农业工程学报,2021,37(9):127-139. DOI: 10.11975/j.issn.1002-6819.2021.09.015.
[26] 白燕英,杨荣花,王会永,等.基于NDVI时序特征的作物样本扩充与遥感精细识别[J].农业机械学报,2025,56(5):370-383. DOI: 10.6041/j.issn.1000-1298.2025.05.035.
[27] Shan Z, Liu Y, Zhou L, et al.ROS-SAM: High-quality interactive segmentation for remote sensing moving object[C]//2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). June 10-17, 2025, Nashville, TN, USA. IEEE, 2025: 3625-3635. DOI: 10.1109/CVPR52734.2025.00343.
[28] Ke L, Ye M, Danelljan M, et al. Segment Anything in High Quality[EB/OL]. arXiv, 2023: 2306.01567. https://arxiv.org/abs/2306.01567.
[29] 张佳华,胡小夏,刘学锋,等.基于MODIS数据提取华北典型区冬小麦种植面积[J].中国科学院大学学报,2013,30(5):637-643. DOI: 10.7523/j.issn.2095-6134.2013.05.010.
[30] Deng K Y, Zhang X Y, Cheng Y J, et al.A remaining useful life prediction method with long-short term feature processing for aircraft engines[J]. Applied Soft Computing, 2020, 93: 106344. DOI: 10.1016/j.asoc.2020.106344.
[31] Sun R, Chen S H, Su H B, et al.The effect of NDVI time series density derived from spatiotemporal fusion of multisource remote sensing data on crop classification accuracy[J]. ISPRS International Journal of Geo-Information, 2019, 8(11): 502. DOI: 10.3390/ijgi8110502.
[32] Mirjalili S, Lewis A.The whale optimization algorithm[J]. Advances in Engineering Software, 2016, 95(C): 51-67. DOI: 10.1016/j.advengsoft.2016.01.008.
[33] Wang D, Qian H Z.CatBoost-based automatic classification study of river network[J]. ISPRS International Journal of Geo-Information, 2023, 12(10): 416. DOI: 10.3390/ijgi12100416.
[34] Strumbelj E, Kononenko I.Explaining prediction models and individual predictions with feature contributions[J]. Knowledge and Information Systems, 2014, 41(3): 647-665. DOI: 10.1007/s10115-013-0679-x.
[35] Ribeiro M, Singh S, Guestrin C. “Why should I trust you?”: Explaining the predictions of any classifier[C]//Proceedings of the2016 Conference of the North American Chapter of the Association for Computational Linguistics: Demonstrations. San Diego, CaliforniaAssociation for Computational Linguistics, 2016: 97-101.10.18653/v1/N16-3020.
[36] Lundberg S M, Lee S I.A unified approach to interpreting model predictions[C]//Proceedings of the 31st International Conference on Neural Information Processing Systems. December 4 - 9, 2017, Long Beach, California, USA. ACM, 2017: 4768-4777. DOI: 10.5555/3295222.3295230.
[37] 刘春红, 李为丽, 焦洁, 等. 一种可解释的云平台任务终止状态预测方法[J]. 计算机研究与发展, 2024, 61(3): 716-727. DOI: 10.7544/issn1000-1239.202220796.
[38] Zhou J L, Diao Q, Liu X, et al.Enhanced garlic crop identification using deep learning edge detection and multi-source feature optimization with random forest[J]. Sensors, 2025, 25(19): 6014. DOI: 10.3390/s25196014.
[39] Yang J, Huang X.The 30 m annual land cover and its dynamics in China from 1990 to 2019[J]. Earth System Science Data Discussions, 2021, 2021: 1-29. DOI: /10.5194/essd-13-3907-2021.
[40] Liu Y F, Chen X H, Chen J, et al.Long-term (2013-2022) mapping of winter wheat in the North China Plain using Landsat data: Classification with optimal zoning strategy[J]. Big Earth Data, 2024, 8(3): 494-521. DOI: 10.1080/20964471.2024.2363552.
[41] 葛建坤, 雷国相, 陈皓锐, 等. 基于SHAP重要性排序和机器学习算法的灌区渠道调度流量预测[J]. 农业工程学报, 2023, 39(13): 113-122. DOI: 10.11975/j.issn.1002-6819.202304081.
[42] 黄青, 杜彦彦. 近20年华北粮食主产区冬小麦种植面积遥感快速提取研究[J]. 中国农业信息, 2023, 35(1): 1-9. DOI: 10.12105/j.issn.1672-0423.20230101.

融合自动样本生成与可解释学习的风云三号时序NDVI冬小麦识别^*

Winter wheat identification from FengYun-3 NDVI time series with automatic sample generation and interpretable learning

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 1

编辑推荐

Metrics

本文评价

访问统计

联系我们