[1]宋宏利,雷海梅,尚明.基于Sentinel 2A/B时序数据的黑龙港流域主要农作物分类[J].江苏农业学报,2021,(01):83-92.[doi:doi:10.3969/j.issn.1000-4440.2021.01.011]
　SONG Hong-li,LEI Hai-mei,SHANG Ming.Crop classification based on Sentinel 2A/B time series data in Heilonggang river basin[J].,2021,(01):83-92.[doi:doi:10.3969/j.issn.1000-4440.2021.01.011]
点击复制

基于Sentinel 2A/B时序数据的黑龙港流域主要农作物分类()
分享到：

江苏农业学报[ISSN:1006-6977/CN:61-1281/TN]

卷:
期数:
2021年01期
页码:
83-92
栏目:
耕作栽培·资源环境
出版日期:
2021-02-28

文章信息/Info

Title:
Crop classification based on Sentinel 2A/B time series data in Heilonggang river basin
作者:
宋宏利; 雷海梅; 尚明
（河北工程大学地球科学与工程学院，河北邯郸056038）
Author(s):
SONG Hong-li; LEI Hai-mei; SHANG Ming
(College of Geosciences and Engineering, Hebei University of Engineering, Handan 056038, China)
关键词:
黑龙港流域; Sentinel 2A/B; 时序植被指数; 农作物分类
Keywords:
Heilonggang river basin; Sentinel 2A/B; time series vegetation index; crop classification
分类号:
TP79
DOI:
doi:10.3969/j.issn.1000-4440.2021.01.011
文献标志码:
A
摘要:
黑龙港流域是中国季节性休耕的试点区域，利用遥感技术及时、准确监测该区域的农作物种植结构，对于精准评价休耕政策的实施效果具有重要意义。本研究以黑龙港流域南部区域为研究对象，采用2019年1-12月的21景Sentinel 2A/B为数据源，构建3种植被指数的时序数据集，结合作物典型时相多光谱数据，采用随机森林法提取研究区内冬小麦、夏玉米、棉花、大蒜、蔬菜等农作物信息，并结合野外调查数据对分类结果进行验证，对比分析不同特征参数提取农作物信息的精度。结果表明，归一化差值红边指数（NDRE1）+典型时相多光谱数据（S20831）取得了最高的总体精度和Kappa系数，其值分别为87.27%和0.85。采用归一化差分植被指数(NDVI)+S20831，冬小麦-夏玉米、大蒜-夏玉米2种双峰型轮作农作物的用户精度最高，分别为94.92%和86.41%。采用S20831+NDRE1，蔬菜及棉花2种单峰型农作物的用户精度最高，分别为91.95%和91.67%。因此，在黑龙港流域农作物的精细分类研究中，结合Sentinel 2A/B植被指数时序数据与典型时相多光谱数据进行分类的精度较高，可用于该区域农作物种植结构监测及休耕政策执行效果评价。
Abstract:
Heilonggang river basin is the pilot area of seasonal fallow in China, using remote sensing technology for timely and accurately monitoring the crop planting structure in this area is of great significance for accurately evaluating the implementation effect of fallow policy. In this study, the southern area of Heilonggang river basin was taken as the research region, and 21 sceneries Sentinel 2A/B from January to December 2019 was used as the data source, the time series data set of three vegetation indices was constructed, combined with the typical time-phase multi spectral data of crops, the winter wheat, summer corn, cotton, garlic, vegetables and other crops in the study area were extracted by random forest classification method. The classified results were verified by combining the above data with the data from field investigation to compare the precision of different characteristic parameters in extracting crop information. The results showed that normalized difference red edge index (NDRE1)+ typical time phase multispectral data (S20831) acquired the best overall accuracy and Kappa coefficient, the value were 87.27% and 0.85, respectively. From the perspective of users, winter wheat-summer maize and garlic-summer maize got the highest accuracy by using normalized difference vegetation index (NDVI)+S2083, the user accuracy was 94.92% and 86.41%, respectively. In addition, vegetables and cotton got the highest accuracy by using S20831 + NDRE1, the user accuracy was 91.95% and 91.67%, respectively. Therefore, the combination of time series Sentinel 2A/B vegetation index and typical time-phase multi-band data has a better effect for the fine classification of crops in Heilonggang river basin, which can be used for the monitoring of crop planting structure and the evaluation of fallow policy effects.

参考文献/References:

[1]杨闫君,占玉林,田庆久,等. 基于GF-1/WFV NDVI时间序列数据的作物分类[J]. 农业工程学报, 2015，31(24): 155-161.
[2]胡琼,吴文斌,宋茜,等. 农作物种植结构遥感提取研究进展[J]. 中国农业科学, 2015, 48(10): 1900-1914.
[3]许文波,田亦陈. 作物种植面积遥感提取方法的研究进展[J]. 云南农业大学学报, 2005, 20(1): 94-98.
[4]LEE E, KASTENS J H, EGBERT S L. Investigating collection 4 versus collection 5 MODIS 250 m NDVI time-series data for crop separability in Kansas, USA[J]. International Journal of Remote Sensing, 2016, 37(1/2): 341-355.
[5]许青云,杨贵军,龙慧灵,等. 基于MODIS NDVI多年时序数据的农作物种植识别[J]. 农业工程学报, 2014，30(11): 134-144.
[6]WARDLOW B D, EGBERT S L. Large-area crop mapping using time-series MODIS 250 m NDVI data: an assessment for the US Central Great Plains[J]. Remote Sensing of Environment, 2008, 112(3): 1096-1116.
[7]SHAO Y, LUNETTA R S, EDIRIWICKREMA J, et al. Mapping cropland and major crop types across the Great Lakes Basin using MODIS-NDVI data[J]. Photogrammetric Engineering & Remote Sensing, 2015, 75(1): 73-84.
[8]OZDOGAN M. The spatial distribution of crop types from MODIS data: temporal unmixing using independent component analysis[J]. Remote Sensing of Environment, 2010, 114(6): 1190-1204.
[9]平跃鹏,臧淑英. 基于MODIS时间序列及物候特征的农作物分类[J]. 自然资源学报, 2016,31(3):503-513．
[10]吴风华,郎婷婷,江东,等. 京津冀地区冬小麦播种面积快速提取技术研究[J]. 江苏农业科学, 2019, 47(16): 224-229.
[11]王文静,张霞,赵银娣,等. 综合多特征的 Landsat 8 时序遥感图像棉花分类方法 [J]. 遥感学报, 2017, 21(1): 115-124．
[12]李晓慧,王宏,李晓兵,等. 基于多时相Landsat 8 OLI影像的农作物遥感分类研究[J].遥感技术与应用, 2019, 34(2): 389-397.
[13]邓刘洋,沈占锋,柯映明,等. 基于地块尺度多时相遥感影像的冬小麦种植面积提取[J].农业工程学报, 2018, 34(21): 165-172.
[14]SRIKANTH P, RAMANA K V, DEEPIKA U, et al. Comparison of various polarimetric decomposition techniques for crop classification[J]. Journal of the Indian Society of Remote Sensing, 2016, 44(4): 635-642.
[15]邱鹏勋,汪小钦,茶明星,等. 基于TWDTW的时间序列GF-1 WFV农作物分类[J]. 中国农业科学, 2019, 52(17): 2951-2961.
[16]黄健熙,侯矞焯,苏伟,等. 基于GF-1 WFV数据的玉米与大豆种植面积提取方法[J]. 农业工程学报, 2017,33(7): 164-170.
[17]刘国栋,邬明权,牛铮,等. 基于GF-1卫星数据的农作物种植面积遥感抽样调查方法[J]. 农业工程学报, 2015, 31(5): 160-166.
[18]刘雅清,王磊,赵希妮,等. 基于GF-1/WFV时间序列的绿洲作物类型提取[J]. 干旱区研究, 2019, 36(3): 781-789.
[19]王冬利,张安兵,赵安周,等. 非监督分类的冬小麦种植信息提取模型[J]. 测绘通报, 2019(8): 68-71.
[20]李冰,梁燕华,李丹丹,等. 多时相GF-1卫星PMS影像提取农作物种植结构[J]. 中国农业资源与区划, 2017,38(9): 56-62.
[21]杜保佳,张晶,王宗明,等. 应用 Sentinel-2A NDVI时间序列和面向对象决策树方法的农作物分类[J]. 地球信息科学学报, 2019, 21(5): 740-751.
[22]郭交,朱琳,靳标. 基于Sentinel-1和Sentinel-2A数据融合的农作物分类[J]. 农业机械学报, 2018, 49(4):192 -198.
[23]吴静,吕玉娜,李纯斌,等. 基于多时相Sentinel-2A的县域农作物分类[J]. 农业机械学报, 2019, 50(9): 194-200.
[24]毕恺艺,牛铮,黄妮,等. 基于Sentinel-2A时序数据和面向对象决策树方法的植被识别[J]. 地理与地理信息科学, 2017, 33(5):16-20.
[25]中央全面深化改革领导小组. 探索实行耕地轮作休耕制度试点方案[R/OL]. (2016-05-20) ［2019-05-20］. https://baike.baidu.com/item/.
[26]FERNNDEZ M A, FERNNDEZ M O, QUINTANO C. Sentinel-2A red-edge spectral indices suitability for discriminating burn severity[J]. International Journal of Applied Earth Observation and Geoinformation, 2016, 50: 170-175.
[27]郑阳,吴炳方,张淼. Sentinel-2数据的冬小麦地上干生物量估算及评价[J]. 遥感学报, 2017, 21(2): 318-328.
[28]TUCKER C J. Red and photographic infrared linear combinations for monitoring vegetation[J]. Remote Sensing and Environment, 1979, 8(2):127-150.
[29]GITELSON A A, GRITZ Y, MERZLYAK M N. Relationships between leaf chlorophyll content and spectral reflectance and algorithms for non-destructive chlorophyll assessment in higher plant leaves[J]. Journal of Plant Physiology, 2003, 160(3): 271-282.
[30]JORDAN C F. Derivation of leaf-area index from quality of light on the forest floor[J]. Ecology, 1969, 50(4): 663-666.
[31]顾娟,李新,黄春林. NDVI时间序列数据集重建方法述评[J]. 遥感技术与应用, 2006, 21(4): 391-395.
[32]BREIMAN L. Random forests[J]. Machine Learning, 2001, 45: 5-32.
[33]刘毅,杜培军,郑辉,等. 基于随机森林的国产小卫星遥感影像分类研究[J]. 测绘科学, 2012, 37(4): 194-196.
[34]JANSSEN L L F, WEL F J M. Accuracy assessment of satellite derived land-cover data: a review[J]. Photogrammetric Engineering & Remote Sensing, 1994, 60(4): 410-432.
[35]LANDIS J R, KOCH G G. The measurement of observer agreement for categorical data[J]. Biometrics, 1977, 33:159-174.

备注/Memo

备注/Memo:
收稿日期：2020-06-01基金项目：河北省自然科学基金项目（D2019402067）；河北省高等学校科学技术研究重点项目（ZD2017212）；河北省研究生示范课建设项目（KCJSX2019065）作者简介：宋宏利（1980-），男，河北抚宁人，博士，副教授，主要从事3S技术集成及应用研究。（E-mail）songholi2003@163.com
更新日期/Last Update: 2021-03-15