| 376 | 2 | 192 |
| 下载次数 | 被引频次 | 阅读次数 |
【目的】新疆奇台县小麦作物种植面积大，秸秆资源丰富。监测并获取秸秆资源的分布与产量，对于促进秸秆综合利用、减少环境污染具有重要意义。【方法】以奇台县境内小麦作物为研究对象，融合多时相的Sentinel-2光学和Sentinel-1雷达数据，选择合适的特征，使用三种机器学习方法对奇台县小麦作物进行识别和小麦秸秆估产。【结果】使用随机森林分类器的分类结果最好，在仅使用Sentinel-1的雷达极化特征分类时，总体精度为96%。在使用Sentinel-1的雷达极化特征与Sentinel-2光学波段特征分类时，总体精度为94%。当使用Sentinel-1的雷达极化特征、Sentinel-2光学波段特征以及植被特征指数的组合分类时精度最高，总体精度可达到98%,Kappa系数可达到97%。基于地面抽样调查数据，估算小麦作物秸秆每公顷产量为27007 kg，总产量为100 615.5 t。【结论】在小麦抽穗期，特征组为Sentinel-1的雷达极化特征、Sentinel-2光学波段特征以及植被特征指数的组合，使用随机森林分类效果最好。
Abstract：【Objective】The wheat crop growing area in Qitai County is extensive,and straw resources are abundant. In order to improve straw consumption and prevent environmental pollution,it is important to detect and obtain the distribution and yield of straw resources.【Methods】This study takes the wheat crop in Qitai County as the research object,integrates the multi-temporal Sentinel-2 optical and Sentinel-1 radar data,selects the relevant features,and adopts three machine learning approaches to identify the wheat crop and estimate the wheat straw yield in Qitai.【Results】The results show that the random forest classifier produced the best classification results,with an overall accuracy of 96% when using only Sentinel-1 radar polarization data for classification. When the combination of Sentinel-1 and Sentinel-2 optical band radar polarisation data was used for classification,the overall accuracy was 94%. With an overall accuracy of 98% and a Kappa coefficient of 97%,the combination of radar polarisation information from Sentinel-1,Sentinel-2 optical band,and vegetation index produced the highest classification accuracy. The predicted production of wheat crop straw was 2 700 kg per hectare,and the totalproduction was 100 616 t based on the ground survey sample data.【Conclusion】During the wheat tassel stage,the random forest classifier with the combination of rader polarisation information from Sentinel-1,Sentinel-2 optical bands,and vegetation index produced the highest classification accuracy.
[1]王志敏，张英华，薛盈文.关于我国小麦生产现状与未来发展的思考[C]第十五次中国小麦栽培科学学术研讨会论文集.北京：中国作物学会，2012:594-605.
[2] He,H. Zhu,X.F. Pan,et al.Study on scale issues in measurement of winter wheat plant area by remote sensing[J]. Remote Sens,2008(1):168–176.
[3] Belgiu,M,L.Drǎgut.Random Forest in Remote Sensing:A Review of Applications and Future Directions[J].Isprs Journal of Photogrammetry and Remote Sensing,2016(114):24-31.
[4]孙建飞，郑聚锋，程琨，等.基于可收集的秸秆资源量估算及利用潜力分析[J].植物营养与肥料学报，2018,24(02):404-413.
[5] Muhammad Sohail Memon,Zhou Jun,Chuanliang Sun.Assessment of Wheat Straw Cover and YieldPerformance in a RiceWheat Cropping System byUsing Landsat Satellite Data[J].Sustainability,2019(11):53-69.
[6] Huawei Mou,Huan Li,Yuguang Zhou,et al.Response of Different Band Combinations in Gaofen-6 WFV for Estimating of Regional Maize Straw Resources Based on RandomForest Classification[J].Sustainability,2021(13):2-16.
[7] S.A.Morain,D.L.Williams.Wheat Production Estimates Using Satellite Images[J]. Agronomy Journal,1975(3):361-364.
[8] R.O.Hoffman,D.M.Edwards,C.C.Eucker.Identifying and measuring crop type using satellite imagery[J].Transactions of the ASAE,1976,19(6):1066-1070.
[9] M. Backes,J. Jacobi. Classification of weed patches in Quickbird images:Verification by ground truth data[J]. EARSeL eProceedings,2006(5):173-179.
[10] Zhang,Y. Ten years of technology advancement in remote sensing and the research in the CRC-AGIP lab in GCE[J].Geomatica,2010(64):173–189.
[11] Kaya G T. A hybrid model for classification of remote sensing images with linear SVM and support vector selection and adaptation[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing,2013,6(4):1988-1997.
[12] Schultz,B. Immitzer,M. Formaggio,et al.Self-guided segmentation and classification of multi-temporal Landsat 8 images for crop type mapping in Southeastern Brazil[J].Remote Sens,2015(7):14482–14508.
[13] Yu T,Douglas S,Hsiang C,et al. Mapping Vegetation and Land Use Types in FanjingshanNational Nature Reserve Using Google Earth Engine[J]. Remote Sensing,2018,10(6):927-941.
[14] Kasper,Johansen,Stuart,et al.Mapping woody vegetation clearing in Queensland,Australiafrom Landsat imagery using the Google Earth Engine[J].Remote Sensing ApplicationsSociety&Environment,2015(1):36-49.
[15] Lee,J.-S. Refifined fifiltering of image noise using local statistics. Comput[J]. Graph. Image Process,1981(15):380–389.
[16] Reigber a,scheiber r,jager m,et al. Very-high-resolution airborne synthetic aperture radar imaging:Signal processing and applications[J]. Proceedings of the IEEE,2012(101):759-783.
[17] Breiman L,Friedman J,Olshen R,et al.Classification and regression trees[M].Chapman&Hall/CRC,1984.
[18] Pal M. Random forest classifier for remote sensing classification[J]. International Journal of Remote Sensing,2005,26(1):217-222.
[19] Immitzer M,Atzberger C,Koukal T. Tree species classification with random forest using very high spatial resolution 8-band WorldView-2 satellite data[J]. Remote Sensing,2012,4(9):2661-2693.
[20]赵萍，傅云飞，郑刘根，等.基于分类回归树分析的遥感影像土地利用/覆被分类研究[J].遥感学报，2005(06):708-716.
[21]何昭欣，张淼，吴炳方，等.Google Earth Engine支持下的江苏省夏收作物遥感提取[J].地球信息科学学报，2019,21(5):126-140.
[22]王利花，金辉虎，王晨丞，等.基于合成孔径雷达的农作物后向散射特性及纹理信息分析—以吉林省农安县为例[J].中国生态农业学报（中英文)，2019,27(9):1385-1393.
基本信息:
DOI：10.16863/j.cnki.1003-6377.2023.05.009
中图分类号:S512.1;TP751
引用信息:
[1]潘竞,艾尼玩·艾买尔,阿斯娅·曼力克,等.基于Sentinel数据的新疆奇台县小麦作物识别和秸秆产量估算[J].草食家畜,2023,No.222(05):51-60.DOI:10.16863/j.cnki.1003-6377.2023.05.009.
基金信息:
新疆维吾尔自治区公益性科研院所基本科研项目“奇台县农作物秸秆产量遥感估算研究”
2023-09-10
2023-09-10