基于时间序列Sentinel-1A数据的玉米种植面积监测研究

doi:10.18402/resci.2018.08.11

摘要/Abstract

摘要：

玉米作为中国三大作物之一,监测其种植面积对及时了解其种植时空分布、保障粮食安全具有重要作用。本文以河北省涿州市为研究区,利用2016年多时相Sentine-1A SAR （Synthetic Aperture Radar, 合成孔径雷达）影像,对玉米种植区域进行提取。在对研究区地物散射特性分析的基础上,分析了微波后向散射特性随不同生育期玉米植株结构发育的变化情况,选择合适时相和极化组合的后向散射系数,运用支持向量机（Support Vector Machine,SVM）算法提取了玉米种植范围和面积信息,并对不同后向散射系数（标准后向散射系数（Sigma-naught,σ₀）和归一化后向散射系数（Gamma-naught,γ₀））用于研究区作物种植区提取的结果进行了比较。结果表明,采用时间序列（4月19日,5月30日,6月11日,7月17日）雷达图像得到的监督分类结果具有较高的分类精度和kappa系数,总体精度达92.96％,Kappa系数为0.91。因此,采用4—7月（春玉米播种至吐丝时期、夏玉米播种至拔节时期）的时间序列SAR数据能有效获取不同种植模式下的玉米信息,而增加8、9月的数据对玉米识别精度的影响不大。总体来说,采用多时相双极化的σ₀数据与相同时相组合的γ₀数据对玉米种植范围提取基本相同,但使用γ₀数据的林地识别精度比σ₀数据提高了3％。研究结果可为多极化SAR数据的玉米识别和面积监测提供参考案例。

关键词: 玉米, Sentinel-1A, SAR, 面积监测, 时间序列, 河北涿州市

Abstract:

Maize is one of the three typical crops in China. Monitoring the distribution of maize is of vital importance in understanding the scope of corn cultivation and ensuring food security. Multi-temporal ESA Sentinel-1A C-band Synthetic Aperture Radar (SAR) VV and VH polarization data at a 20 m spatial resolution were used to identify maize identity for Zhuozhou City in Hebei province, China. Based on the analysis of the scattering characteristics for different crops, as well as the backscattering coefficients variability of different crop structures at different growth and development stages of the study area, dual-polarized backscatter coefficients from appropriate phases were selected. To assess the performance of different backscattering coefficients forms in classification, both σ₀ and γ₀ data were used. Compared with the sampling data from the field campaign, the results showed that a high recognition accuracy of maize can be obtained using support vector machine (SVM) method if time series data are used. Classification result obtained from time series radar images has a higher accuracy and kappa coefficient than those from a single image. For example, results gotten from long time series σ₀ data have the ideal performance with overall accuracy of 92.96% and a Kappa coefficient of 0.91. Data from sowing to silking of spring maize and sowing to jointing of summer maize (April 19, May 30, June 11, and July 17) can effectively obtain maize information under different cropping patterns. The increase data of August and September would have little impact on the accuracy of maize identification. The differences between σ₀ and γ₀ data are relatively moderate. However, the classification accuracy obtained with long time series γ₀ data can be improved by 3% than that with σ₀ data for the forest area.

Key words: maize, Sentinel-1A, SAR, area monitoring, time-series, Heibei Zhuozhou

李俐, 孔庆玲, 王鹏新, 王蕾, 荀兰. 基于时间序列Sentinel-1A数据的玉米种植面积监测研究[J]. 资源科学, 2018, 40(8): 1608-1621.

Li LI, Qingling KONG, Pengxin WANG, Lei WANG, Lan XUN. Monitoring of maize planting area based on time-series Sentinel-1A SAR data[J]. Resources Science, 2018, 40(8): 1608-1621.

图/表 16

图1

表1

表2

图2

图3

图4

图5

图6

表3

表4

图7

图8

表5

表6

图9

表7

参考文献 30

[1]	赵久然, 王荣焕, 刘新香. 我国玉米产业现状及生物育种发展趋势[J]. 生物产业技术, 2016, (3): 45-52.
	[Zhao J R, Wang R H, Liu X X.Current situation of China’s corn industry and biological breeding development trend[J]. Biotechnology & Business, 2016, (3): 45-52. ]
[2]	钱永兰, 杨邦杰, 焦险峰. 基于遥感抽样的国家尺度农作物面积统计方法评估[J]. 农业工程学报, 2007, 23(11): 180-187.
	[Qian Y L, Yang B J, Jiao X F.Accuracy assessment on the crop area estimating method based on RS sampling at national scale[J]. Transactions of the Chinese Society of Agricultural Engineering, 2007, 23(11): 180-187. ]
[3]	陈水森, 柳钦火, 陈良富, 等. 粮食作物播种面积遥感监测研究进展[J]. 农业工程学报, 2005, 21(6): 166-170.
	[Chen S S, Liu Q H, Chen L F, et al. Review of research advances in remote sensing monitoring of grain crop area[J]. Transactions of the Chinese Society of Agricultural Engineering, 2005, 21(6): 166-170. ]
[4]	唐华俊, 吴文斌, 杨鹏, 等. 作物空间格局遥感监测研究进展[J]. 中国农业科学, 2010, 43(14): 2879-2888.
	[Tang H J, Wu W B, Yang P, et al. Recent progresses in monitoring crop spatial patterns by using remote sensing technologies[J]. Scientia Agricultura Sinica, 2010, 43(14): 2879-2888. ]
[5]	王迪, 周清波, 陈仲新, 等. 基于合成孔径雷达的农作物识别研究进展[J]. 农业工程学报, 2014, 30(16): 203-212.
	[Wang D, Zhou Q B, Chen Z X, et al. Research advances on crop identification using synthetic aperture radar[J]. Transactions of the Chinese Society of Agricultural Engineering, 2014, 30(16): 203-212. ]
[6]	Frate F D, Schiavon G, Solimini D, et al. Crop classification using multi-configuration C-band SAR data[J]. IEEE Transactions on Geoscience & Remote Sensing, 2003, 41(7): 1611-1619.
[7]	Zeyada H H, Ezz M M, Nasr A H, et al. Evaluation of the discrimination capability of full polarimetric SAR data for crop classification[J]. International Journal of Remote Sensing, 2016, 37(11): 2585-2603.
[8]	Forkuor G, Conrad C, Thiel M, et al. Integration of optical and synthetic aperture radar imagery for improving crop mapping in Northwestern Benin, West Africa[J]. Remote Sensing, 2014, 6(7): 6472-6499.
[9]	Sukawattanavijit C, Chen J.Fusion of Multi-Frequency SAR Data with THAICHOTE Optical Imagery for Maize Classification in Thailand[C]. Milan: Geoscience and Remote Sensing Symposium, 2015.
[10]	东朝霞, 王迪, 周清波, 等. 基于SAR遥感的北方旱地秋收作物识别研究[J]. 中国农业资源与区划, 2016, 37(8): 27-36.
	[Dong Z X, Wang D, Zhou Q B, et al. Dryland crop identification based on synthetic aperture radar in the north china plain[J]. Chinese Journal of Agricultural Resource and Regional Planning, 2016, 37(8): 27-36. ]
[11]	Nguyen D B, Gruber A, Wagner W.Mapping rice extent and cropping scheme in the Mekong Delta using Sentinel-1A data[J]. Remote Sensing Letters, 2016, 7(12): 1209-1218.
[12]	Chen C F, Son N T, Chen C R, et al. Rice Crop Mapping using Sentinel-1A Phenological Metrics[C]. Prague: International Archives of the Photogrammetry, 2016.
[13]	Abdikan S, Sanli F B, Ustuner M, et al. Land Cover Mapping Using Sentinel-1 SAR Data[C]. Prague: International Archives of the Photogrammetry, 2016.
[14]	Sonobe R, Yamaya Y, Tani H, et al.Assessing the suitability of data from Sentinel-1A and 2A for crop classify-cation[J]. Giscience & Remote Sensing, 2017, 54(6): 1-21.
[15]	Torres R, Snoeij P, Geudtner D, et al. GMES Sentinel-1 mission[J]. Remote Sensing of Environment, 2012, 120(6): 9-24.
[16]	雷万明, 刘光炎, 黄顺吉. 分布式卫星SAR的波束形成和多视处理成像[J]. 电子与信息学报, 2002, 24(11): 1620-1626.
	[Lei W M, Liu G Y, Huang S J.The imaging of distributed satellite SAR system using beamforming and multi look[J]. Journal of Electronics and Information Technology, 2002, 24(11): 1620-1626. ]
[17]	Lee J S, Grunes M R, Grandi G D.Polarimetric SAR speckle filtering and its implication for classification[J]. IEEE Transactions on Geoscience & Remote Sensing, 1999, 37(5): 2363-2373.
[18]	杜伟娜, 徐爱功, 宋耀鑫, 等. 新型SAR传感器一级地距产品绝对辐射定标方法[J]. 国土资源遥感, 2016, 28(4): 30-34.
	[Du W N, Xu A G, Song Y X, et al. Absolute radiometric calibration of level-1 detected ground range products of new SAR sensors[J]. Remote Sensing for Land & Resources, 2016, 28(4): 30-34. ]
[19]	Small D.Flattening gamma: radiometric terrain correction for SAR imagery[J]. IEEE Transactions on Geoscience & Remote Sensing, 2011, 49(8): 3081-3093.
[20]	Miettinen J, Liew S C.Separability of insular Southeast Asian woody plantation species in the 50 m resolution ALOS PALSAR mosaic product[J]. Remote Sensing Letters, 2010, 2(4): 299-307.
[21]	Wu F, Wang C, Zhang H, et al. Rice crop monitoring in South China With RADARSAT-2 quad-polarization SAR data[J]. IEEE Geoscience & Remote Sensing Letters, 2011, 8(2): 196-200.
[22]	贾坤, 李强子, 田亦陈, 等. 遥感影像分类方法研究进展[J]. 光谱学与光谱分析, 2011, 31(10): 2618-2623.
	[Jia K, Li Q Z, Tian Y C, et al. A review of classification methods of remote sensing imagery[J]. Spectroscopy and Spectral Analysis, 2011, 31(10): 2618-2623. ]
[23]	王强, 孙洪. 基于支持向量机的多极化SAR图像监督分类[J]. 信号处理, 2005, 21(Z1): 531-534.
	[Wang Q, Sun H.Supervised classification of polarimetric SAR image based on support vector machine[J]. Signal Processing, 2005, 21(Z1): 531-534. ]
[24]	Loosvelt L, Peters J, Skriver H, et al. Random forests as a tool for estimating uncertainty at pixel-level in SAR image classification[J]. International Journal of Applied Earth Observations & Geoinformation, 2012, 19(1): 173-184.
[25]	Erener A.Classification method, spectral diversity, band combination and accuracy assessment evaluation for urban feature detection[J]. International Journal of Applied Earth Observations & Geoinformation, 2013, 21(4): 397-408.
[26]	Schwert B, Rogan J, Giner N M, et al. A comparison of support vector machines and manual change detection for land-cover map updating in Massachusetts, USA[J]. Remote Sensing Letters, 2013, 4(9): 882-890.
[27]	Dixon B, Candade A.Multispectral land use classification using neural networks and support vector machines: one or the other, or both?[J]. International Journal of Remote Sensing, 2008, 29(4): 1185-1206.
[28]	Petropoulos G P, Kalaitzidis C, Vadrevu P K.Support vector machines and object-based classification for obtaining land-use/cover cartography from Hyperion hyperspectral imagery[J]. Computers & Geosciences, 2012, 41(2): 99-107.
[29]	汤井田, 胡丹, 龚智敏. 基于SVM的SAR图像分类研究[J]. 遥感技术与应用, 2008, 23(3): 341-345.
	[Tang J T, Hu D, Gong Z M.Study of classification by support vector machine on synthetic aperture radar image[J]. Remote Sensing Technology and Application, 2008, 23(3): 341-345. ]
[30]	Richards J A.Remote Sensing Digital Image Analysis[M]. Berlin: Springer-Verlag, 1999.

	四月			五月			六月			七月			八月			九月			十月
	上旬	中旬	下旬	上旬	中旬	下旬	上旬	中旬	下旬	上旬	中旬	下旬	上旬	中旬	下旬	上旬	中旬	下旬	上旬	中旬	下旬
春玉米	播种、出苗			三叶、七叶			拔节			抽雄、吐丝			乳熟、成熟收获			收获
夏玉米							播种、三叶			七叶、拔节			吐丝			乳熟、成熟			收获

序号	获取时间	不同种植模式下玉米生长发育期
序号	获取时间	单季夏玉米	轮作夏玉米	春玉米
D1	2016-04-19	裸地或豌豆*	冬小麦*	播种
D2	2016-05-30	裸地*	冬小麦*	七叶
D3	2016-06-11	播种	播种	拔节前期
D4	2016-07-17	拔节	拔节	吐丝
D5	2016-08-10	拔节吐丝	吐丝	乳熟成熟
D6	2016-09-03	乳熟	乳熟	收获

J-M距离	4月单时相冬小麦数据	5月单时相冬小麦数据	6月单时相轮作夏玉米数据	7月单时相轮作夏玉米数据	8月单时相轮作夏玉米数据	9月单时相轮作夏玉米数据	4—7月时间序列轮作夏玉米数据
建筑	1.47	1.76	1.49	1.56	1.68	1.60	1.99
水体	1.58	1.92	1.79	1.77	1.98	1.90	1.96
树林	0.61	0.72	0.69	0.68	0.70	0.81	1.86
裸地	0.72	0.69	0.68	0.70	0.81	1.12	1.85
春玉米	0.62	1.72	1.46	0.32	0.20	0.09	1.93
单季夏玉米	0.60	1.85	0.97	0.26	0.24	0.14	1.94
其它	0. 68	1.36	1.71	0.42	0.36	0.62	1.90

J-M距离	4月单时相冬小麦数据	5月单时相冬小麦数据	6月单时相轮作夏玉米数据	7月单时相轮作夏玉米数据	8月单时相轮作夏玉米数据	9月单时相轮作夏玉米数据	4-7月时间序列轮作夏玉米数据
建筑	1.56	1.48	1.39	1.72	1.48	1.58	1.96
水体	1.86	1.86	1.78	1.91	1.87	1.86	1.98
树林	1.23	1.34	1.34	0.86	1.23	1.15	1.94
裸地	0.72	0.63	0.64	0.76	0.70	0.78	1.82
春玉米	0.59	1.63	1.46	0.32	0.20	0.09	1.86
单季夏玉米	0.51	1.78	0.86	0.04	0.21	0.10	1.84
其它	0.42	0.65	0.51	0.48	0.62	0.76	1.89

类别	时间序列SAR数据精度验证					单时相SAR数据精度验证
	采用σ₀分类		采用γ₀分类			采用σ₀分类		采用γ₀分类
	用户精度	制图精度	用户精度	制图精度	用户精度		制图精度	用户精度	制图精度
轮作夏玉米	95.56	94.85	96.40	94.81		0.23	0.20	10.85	10.24
建筑	99.34	94.36	96.34	99.62		97.83	89.40	99.31	94.70
林地	89.54	92.70	92.15	96.42		62.07	15.65	88.78	79.13
春玉米	86.67	91.04	78.35	90.24		81.69	89.68	89.20	93.92
单季夏玉米	92.81	90.88	92.71	78.30		7.27	2.86	3.94	2.62
水	98.11	99.36	100.00	95.48		93.68	83.44	79.59	90.40
裸地	92.91	88.72	97.24	86.12		14.79	39.17	5.90	11.06
其它	90.91	88.46	76.67	87.62		47.48	70.75	90.95	75.49
总体精度	92.96		91.67			45.15		52.81
Kappa系数	0.92		0.88			0.37		0.46