欢迎访问《草业学报》官方网站,今天是 分享到:

草业学报 ›› 2016, Vol. 25 ›› Issue (7): 1-12.DOI: 10.11686/cyxb2015433

• 研究论文 •    下一篇

基于Landsat 8 OLI和MODIS数据的高寒草地盖度升尺度效应研究——以夏河县桑科草原试验区为

孟宝平1, 崔霞2, 杨淑霞1, 高金龙1, 胡远宁1, 陈思宇1, 梁天刚1, *   

  1. 1.草地农业生态系统国家重点实验室,兰州大学草地农业科技学院,甘肃 兰州 730020;
    2.兰州大学 西部环境教育部重点实验室,兰州大学资源与环境学院,甘肃 兰州 730000
  • 收稿日期:2015-09-14 修回日期:2015-11-16 出版日期:2016-07-20 发布日期:2016-07-20
  • 通讯作者: E-mail:tgliang@lzu.edu.cn
  • 作者简介:孟宝平(1989),男,甘肃陇西人,在读博士。E-mail:mengbp09@lzu.edu.cn
  • 基金资助:

    国家自然科学基金项目(31372367,31228021,41401472),农业部公益性行业(农业)科研专项项目(201203006)和长江学者和创新团队发展计划(IRT13019)资助

Scaling-up methodology for alpine grassland coverage monitoring based on Landsat 8 OLI and MODIS remote sensing data: A case study in XiaheSangke grassla

MENG Bao-Ping1, CUI Xia2, YANG Shu-Xia1, GAO Jin-Long1, HU Yuan-Ning1, CHEN Si-Yu1, LIANG Tian-Gang1, *   

  1. 1.State Key Laboratory of Grassland Agro-ecosystems, College of Pastoral Agriculture Science and Technology, Lanzhou University, Lanzhou 730020, China;
    2.Key Laboratory of Western China’s Environmental Systems(Ministry of Education), College of Earth and Environmental Sciences, Lanzhou University, Lanzhou 730000, China;
  • Received:2015-09-14 Revised:2015-11-16 Online:2016-07-20 Published:2016-07-20

摘要:

基于Landsat 8 OLI和MODIS卫星遥感资料,结合2013-2014年甘南州夏河县桑科草原试验区野外实测盖度数据,对草地盖度敏感的OLI波段及其组合指数进行了筛选,构建并确立了试验区基于MODIS植被指数的草地盖度反演模型,探讨了不同空间升尺度方式对草地盖度的空间效应。结果表明,1)OLI比值植被指数r(Band7/Band5)对草地盖度的响应最敏感,基于该指数的草地盖度最优反演模型为yoli=-270.064xoli+115.987(R2=0.833,P<0.001);2)基于MODIS MEVI和Landsat 8 OLI比值植被指数r(30 m)反演的盖度重采样数据(250 m)的对数模型为最优草地盖度评估模型(R2=0.795,P<0.001),其决定系数较MODIS MEVI与基于农业多光谱照相机(agricultural digital camera,ADC)的盖度数据建立的回归模型高(R2=0.706),绝对误差和相对误差较低;3)基于OLI和MODIS植被指数的3种草地盖度回归模型(方法1~3)精度均优于直接使用MODIS植被指数建立的回归模型(方法4);方法3利用OLI 比值指数r反演盖度(30 m),将其升尺度至250 m,再反演盖度构建的对数模型的精度最高(R2=0.795),其次依次为方法2构建的模型(R2=0.760)、方法1构建的模型(R2=0.730)和方法4构建的模型(R2=0.706)。

Abstract:

This research used remote sensing data of MODIS and Landsat 8 OLI, combined with ground observations during 2013 and 2014 in XiaheSangke grassland, Gansu Province. Both individual bands and combinations of bands of Landsat 8 OLI were tested, with a view to selecting band combinations sensitive to grassland coverage. Then, grassland coverage inversion models were established based on MODIS vegetation index data. At the same time, the spatial scale effect was analyzed with a 30 m resolution and up-scaled to 250 m resolution for spectral reflectance, vegetation index, and estimated grassland coverage. It was found that: 1) the ratio of Band7/Band5 of the OLI data was the most sensitive combination for detecting grassland coverage, and the best grassland coverage inversion model was the linear function: yoli=-270.064xoli+115.987, R2=0.833, P<0.001; 2) The best grassland coverage inversion model was the logarithmic model (y=64.160ln(xMEVI)+136.927, R2=0.795, P<0.001), which was established by using MODIS MEVI and up-scaling ratio index of OLI. The coefficient of fit was higher than for the models based on MODIS MEVI and Agricultural Digital Camera pictures (R2=0.706), and its average absolute error and average relative error were lower. 3) The accuracy of the logarithmic model (R2=0.795) based on MODIS MEVI and up-scaling grassland coverage using a ratio index (Band7/Band5) of OLI was higher than other models.