基于Sentinel-2的青铜峡灌区水稻和玉米种植分布早期识别

doi:10.12118/j.issn.1000-6060.2023.541

摘要/Abstract

摘要：

及时准确地掌握灌区内作物种植分布对于灌溉水资源高效配置、农田精准管理具有重要指导意义。以宁夏青铜峡灌区为研究对象，利用多时相Sentinel-2卫星数据，通过水稻和玉米早期特征分析，提取关键的“水淹”信号和“植被”信号，构建时序归一化差异水体指数（MNDWI）和归一化植被指数（NDVI）特征值数据集，并通过样本分析关键特征阈值，构建水稻和玉米早期种植分布决策树模型，提取2022年宁夏青铜峡灌区水稻和玉米种植的空间分布。结果表明:（1）玉米和水稻苗期的后半段5月15—31日，水淹信号和植被信号是区分二者关键时期。（2）基于早期作物物候特征的方法，在5月16—31日获取的水稻和玉米图像制图精度高于90%，用户精度超过91%，总体精度超过90%，Kappa系数高于0.88，明显高于同时期随机森林方法的分类精度。（3）本研究提出的方法在早期水稻和玉米种植分布提取方面具有较强的适用性，并且能够在时空尺度上以较少的实地样本进行延展，同时在时间上也更有优势。因此，该方法为青铜峡灌区水稻和玉米种植分布早期调查提供了重要的方法支撑。

关键词: 青铜峡灌区, Sentinel-2, 归一化植被指数, 归一化差异水体指数, 决策树, 水稻, 玉米

Abstract:

Timely and accurate understanding of crop distribution within irrigation areas is essential for the efficient allocation of irrigation water resources and precise field management. This study focuses on the Qingtongxia irrigation area in Ningxia, China, employing multitemporal Sentinel-2 satellite data to analyze early characteristics of rice and maize. Key “flooding” and “vegetation” signals are extracted, and a time-series dataset comprising the modified normalized difference water index (MNDWI) and normalized vegetation index (NDVI) is constructed. By analyzing sample thresholds for these key features, a decision tree model for the early planting distribution of rice and maize is established, facilitating the extraction of the spatial distribution for rice and maize planting in the Qingtongxia irrigation area in 2022. The results reveal the following: (1) During the latter half of the maize and rice seedling stages, from May 15 to 31, flooding and vegetation signals are crucial for differentiating between the two crops. (2) Based on the early crop phenological characteristics, the mapping accuracy of rice and corn images obtained from May 16 to May 31 was higher than 90%, with user accuracy exceeding 91% and overall accuracy exceeding 90%. The Kappa coefficient was higher than 0.88, significantly higher than the classification accuracy of the random forest classification method during the same period. (3) The proposed method demonstrates strong applicability in the early extraction of rice and maize planting distribution, requiring fewer ground samples for extension across both spatial and temporal scales. Therefore, this method provides significant support for early investigations of rice and maize planting distribution in the Qingtongxia irrigation area.

Key words: Qingtongxia irrigation area, Sentinel-2, normalized vegetation index, normalized difference water index, decision tree, rice, maize

朱磊, 王科, 丁一民, 孙振源, 孙伯颜. 基于Sentinel-2的青铜峡灌区水稻和玉米种植分布早期识别[J]. 干旱区地理, 2024, 47(5): 850-860.

ZHU Lei, WANG Ke, DING Yimin, SUN Zhenyuan, SUN Boyan. Early identification of rice and corn planting distribution in Qingtongxia irrigation area based on Sentinel-2[J]. Arid Land Geography, 2024, 47(5): 850-860.

图/表 12

表1

图1

表2

图2

图3

图4

图5

表3

图6

表4

图7

表5

参考文献 29

[1]	孙智虎, 张锦水, 洪友堂, 等. GF-7卫星多角度特征作物识别[J]. 遥感学报, 2023, 27(9): 2127-2138.
	[Sun Zhihu, Zhang Jinshui, Hong Youtang, et al. Crop recognition by multiangle features of GF-7 satellite[J]. National Remote Sensing Bulletin, 2023, 27(9): 2127-2138. ]
[2]	田艳君, 石莹, 帅艳民, 等. 基于遥感时序特征的地表覆被信息提取[J]. 干旱区地理, 2021, 44(2): 450-459.
	[Tian Yanjun, Shi Ying, Shuai Yanmin, et al. Land cover information retrieval from temporal features based remote sensing images[J]. Arid Land Geography, 2021, 44(2): 450-459. ]
[3]	白燕英, 高聚林, 张宝林. 基于Landsat8影像时间序列NDVI的作物种植结构提取[J]. 干旱区地理, 2019, 42(4): 893-901.
	[Bai Yanying, Gao Julin, Zhang Baolin. Extraction of crop planting structure based on time-series NDVI of Landsat8 images[J]. Arid Land Geography, 2019, 42(4): 893-901. ]
[4]	欧阳玲, 毛德华, 王宗明, 等. 基于GF-1与Landsat8 OLI影像的作物种植结构与产量分析[J]. 农业工程学报, 2017, 33(11): 147-156.
	[Ou Yangling, Mao Dehua, Wang Zongming, et al. Analysis crops planting structure and yield based on GF-1 and Landsat8 OLI images[J]. Transactions of the Chinese Society of Agricultural Engineering, 2017, 33(11): 147-156. ]
[5]	刘焕军, 孟令华, 邱政超, 等. 棉花生长初期灌溉信息遥感提取与校正[J]. 中国生态农业学报, 2017, 25(8): 1216-1223.
	[Liu Huanjun, Meng Linghua, Qiu Zhengchao, et al. Using remote sensing to extract and correct irrigation data during early cotton growth stage[J]. Chinese Journal of Eco-Agriculture, 2017, 25(8): 1216-1223. ]
[6]	尹瀚民, 古丽·加帕尔, 于涛, 等. 哈萨克斯坦北部小麦遥感估产方法研究[J]. 干旱区地理, 2022, 45(2): 488-498.
	[Yin Hanmin, Jiapaer Guli, Yu Tao, et al. Research on remote sensing estimation of wheat yield in northern Kazakhstan[J]. Arid Land Geography, 2022, 45(2): 488-498. ]
[7]	郝鹏宇. 基于多时相遥感数据的农作物分类研究[D]. 北京: 中国科学院大学, 2017.
	[Hao Pengyu. Crop classification using time series remote sensing data[D]. Beijing: University of the Chinese Academy of Sciences, 2017. ]
[8]	郝鹏宇, 唐华俊, 陈仲新, 等. 基于历史增强型植被指数时序的农作物类型早期识别[J]. 农业工程学报, 2018, 34(13): 179-186.
	[Hao Pengyu, Tang Huajun, Chen Zhongxin, et al. Early season crop type recognition based on historical EVI time series[J]. Transactions of the Chinese Society of Agricultural Engineering, 2018, 34(13): 179-186. ]
[9]	徐美, 阮本清, 黄诗峰, 等. 灌区作物种植结构遥感监测及其应用[J]. 水利学报, 2007, 38(7): 879-885.
	[Xu Mei, Ruan Benqing, Huang Shifeng, et al. Monitoring of crop variety distribution by remote sensing and its application[J]. Journal of Hydraulic Engineering, 2007, 38(7): 879-885. ]
[10]	Cai Y P, Guan K Y, Peng J, et al. A high-performance and in-season classification system of field-level crop types using time-series Landsat data and a machine learning approach[J]. Remote Sensing of Environment, 2018, 210: 35-47.
[11]	Zhou F Q, Zhang A N, Townley-Smith L. A data mining approach for evaluation of optimal time-series of MODIS data for land cover mapping at a regional level[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2013, 84: 114-129.
[12]	Villa P, Stroppiana D, Fontanelli G, et al. In-season mapping of crop type with optical and X-Band SAR data: A classification tree approach using synoptic seasonal features[J]. Remote Sensing, 2015, 7(10): 12859-12886.
[13]	Azar R, Villa P, Stroppiana D, et al. Assessing in-season crop classification performance using satellite data: A test case in northern Italy[J]. European Journal of Remote Sensing, 2016, 49(1): 361-380.
[14]	杨丽慧. 青铜峡灌区农业节水的农田水土环境响应研究[D]. 北京: 中国水利水电科学研究院, 2017.
	[Yang Lihui. Research on farmland water and soil environment response to water saving in Qingtongxia irrigation district[D]. Beijing: China Academy of Water Resources and Hydropower Research, 2017. ]
[15]	Weinmann M, Maier P M, Florath J, et al. Investigations on the potential of hyperspectral and Sentinel-2 data for land-cover/land-use classification[J]. ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, 2018, 4: 155-162.
[16]	Xia T, Ji W W, Li W D, et al. Phenology-based decision tree classification of rice-crayfish fields from Sentinel-2 imagery in Qianjiang, China[J]. International Journal of Remote Sensing, 2021, 42(21): 8124-8144.
[17]	苏伟, 张明政, 蒋坤萍, 等. Sentinel-2卫星影像的大气校正方法[J]. 光学学报, 2018, 38(1): 322-331.
	[Su Wei, Zhang Mingzheng, Jiang Kunping, et al. Atmospheric correction method for Sentinel-2 Satellite imagery[J]. Acta Optica Sinica, 2018, 38(1): 322-331. ]
[18]	张悦琦, 李荣平, 穆西晗, 等. 基于多时相GF-6遥感影像的水稻种植面积提取[J]. 农业工程学报, 2021, 37(17): 189-196.
	[Zhang Yueqi, Li Rongping, Mu Xihan, et al. Extraction of paddy rice planting areas based on multi-temporal GF-6 remote sensing images[J]. Transactions of the Chinese Society of Agricultural Engineering, 2021, 37(17): 189-196. ]
[19]	魏浩东, 杨靖雅, 蔡志文, 等. 物候窗口和多源中高分辨率影像的稻虾田提取[J]. 遥感学报, 2022, 26(7): 1423-1436.
	[Wei Haodong, Yang Jingya, Cai Zhiwen, et al. Extraction of rice and shrimp fields from phenological windows and multi-source high-resolution images[J]. Journal of Remote Sensing, 2022, 26(7): 1423-1436. ]
[20]	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, 9(5): 589-595.
	[Xu Hanqiu. A study on information extraction of water body with the modified normalized difference water index (MNDWI)[J]. National Remote Sensing Bulletin, 2005, 9(5): 589-595. ]
[21]	姜伊兰, 陈保旺, 黄玉芳, 等. 基于Google Earth Engine和NDVI时序差异指数的作物种植区提取[J]. 地球信息科学学报, 2021, 23(5): 938-947. doi: 10.12082/dqxxkx.2021.200291
	[Jiang Yilan, Chen Baowang, Huang Yu fang, et al. Crop planting area extraction based on Google Earth Engine and NDVI time series difference index[J]. Journal of Geo-information Science, 2021, 23(5): 938-947. ]
[22]	Breiman L. Random forests[J]. Machine Learning, 2001(45): 5-32.
[23]	杨庆振, 郭敏, 范新成. 基于随机森林算法的高光谱遥感作物分类[J]. 测绘与空间地理信息, 2023, 46(4): 149-151, 154.
	[Yang Qingzhen, Guo Min, Fan Xincheng. Hyperspectral remote sensing crop classification based on random forest algorithm[J]. Surveying and Spatial Geographic Information, 2023, 46(4): 149-151, 154. ]
[24]	解毅, 王佳楠, 刘钰. 基于Sentinel-1/2数据特征优选的冬小麦种植区识别方法研究[J]. 农业机械学报, 2024, 55(2): 231-241.
	[Xie Yi, Wang Jianan, Liu Yu. Winter wheat identification based on feature optimization of Sentinel-1/2 data[J]. Transactions of the Chinese Society for Agricultural Machinery, 2024, 55(2): 231-241. ]
[25]	van der Linden S, Rabe A, Held M, et al. The EnMAP-Box: A toolbox and application programming interface for EnMAP data processing[J]. Remote Sensing, 2015, 7(9): 11249-11266.
[26]	张馨予, 蔡志文, 杨靖雅, 等. 时序滤波对农作物遥感识别的影响[J]. 农业工程学报, 2022, 38(4): 215-224.
	[Zhang Xinyu, Cai Zhiwen, Yang Jingya, et al. Impacts of temporal smoothing methods on crop type identification[J]. Transactions of the Chinese Society of Agricultural Engineering, 2022, 38(4): 215-224. ]
[27]	Congalton R G. A review of assessing the accuracy of classifications of remotely sensed data[J]. Remote Sensing of Environment, 1991, 37(1): 35-46.
[28]	Hay A M. The derivation of global estimates from a confusion matrix[J]. International Journal of Remote Sensing, 1988, 9(8): 1395-1398.
[29]	You N S, Dong J W, Huang J J X, et al. The 10-m crop type maps in northeast China during 2017—2019[J]. Scientific Data, 2021, 8(1): 41.

月份	2	3			4			5			6			7			8			9
旬	下	上	中	下	上	中	下	上	中	下	上	中	下	上	中	下	上	中	下	上	中	下
水稻								播种		出苗	分蘖		拔节			抽穗			乳熟	成熟
玉米						播种				出苗	拔节			抽雄		开花		灌浆		成熟
小麦	播种		出苗		三叶	分蘖		拔节		抽穗	开花		灌浆	成熟
葡萄				展藤	萌芽	出叶				开花	坐果			膨大			转色			成熟
枸杞				萌芽	展叶			开花		幼果			成熟

生育期	获取时间（年-月-日）	传感器与数据级别	数量/景	云覆盖率/%
播种前	2022-01-04	S2A-MSIL2A	6	<8
播种前	2022-03-15	S2A-MSIL2A	6	<5
播种期	2022-04-06	S2B-MSIL2A	5	<5
	2022-04-24	S2A-MSIL2A	5	<8
	2022-05-01	S2A-MSIL2A	6	<8
	2022-05-09	S2B-MSIL2A	6	<5
出苗期	2022-05-16	S2B-MSIL2A	5	<8
	2022-05-24	S2A-MSIL2A	6	<8
	2022-05-31	S2A-MSIL2A	6	<5
拔节期	2022-06-05	S2B-MSIL2A	6	<5
	2022-06-13	S2A-MSIL2A	6	<5
	2022-06-15	S2B-MSIL2A	6	<5
	2022-06-30	S2A-MSIL2A	6	<5

生育期	日期	玉米		水稻		总体精度/%	Kappa系数
生育期	日期	制图精度/%	用户精度/%	制图精度/%	用户精度/%	总体精度/%	Kappa系数
播种期	5月1日—16日	95.62	94.33	93.02	92.66	97.0413	0.920
出苗期	5月1日—24日	95.25	93.31	94.87	93.34	95.4182	0.908
	5月1日—31日	93.49	92.87	93.54	91.56	91.5336	0.882
拔节期	5月1日—6月05日	90.96	91.41	92.06	93.14	90.6971	0.851
拔节期	5月1日—6月13日	89.36	90.61	90.84	91.33	89.5663	0.801
	5月1日—6月30日	63.53	92.86	95.87	75.69	81.0671	0.609

生育期	日期	玉米		水稻		总体精度/%	Kappa系数
生育期	日期	制图精度/%	用户精度/%	制图精度/%	用户精度/%	总体精度/%	Kappa系数
播种期	5月16日	89.34	87.56	90.01	89.35	90.01	0.89
出苗期	5月24日	89.98	90.20	89.97	90.02	90.06	0.87
	5月31日	87.34	88.75	86.37	89.23	89.09	0.84
拔节期	6月05日	90.02	90.51	90.07	91.24	90.76	0.90
	6月13日	88.09	90.03	89.67	90.01	89.12	0.85
	6月30日	91.03	90.5	86.08	87.09	86.57	0.86

方法	最大值	最大值日期（年-月-日）	Kappa系数≥0.9 的首个日期（年-月-日）
基于早期物候特征方法	0.92	2022-05-16	2022-05-16
随机森林方法	0.90	2022-06-05	2022-06-05