草地学报 ›› 2014, Vol. 22 ›› Issue (4): 706-712.DOI: 10.11733/j.issn.1007-0435.2014.04.005

• 研究论文 • 上一篇    下一篇

基于高分遥感数据校正的MODIS草地生物量监测模型精度评价

胡远宁, 冯琦胜, 陈思宇, 修丽娜, 梁天刚   

  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
  • 收稿日期:2013-11-05 修回日期:2014-01-24 出版日期:2014-08-15 发布日期:2014-08-04
  • 通讯作者: 梁天刚
  • 作者简介:胡远宁(1991-),女,甘肃定西人,硕士研究生,研究方向为草地遥感与地理信息系统,E-mail:huyn2012@lzu.cn
  • 基金资助:

    国家自然科学基金项目(31228021);农业部公益性行业(农业)科研专项项目(201203006);中国工程院重大咨询研究项目(2012-ZD-7);长江学者和创新团队发展计划(IRT13019);兰州大学中央高校基本科研业务费专项资金(lzujbky-2013-212)资助

Accuracy Evaluation of the MODIS Grassland Biomass Monitoring Model Based on the Correction of High-resolution Remote Sensing Data

HU Yuan-ning, FENG Qi-sheng, CHEN Si-yu, XIU Li-na, LIANG Tian-gang   

  1. State Key Laboratory of Grassland Agro-ecosystems, College of Pastoral Agriculture Science and Technology, Lanzhou University, Lanzhou, Gansu Province 730020, China
  • Received:2013-11-05 Revised:2014-01-24 Online:2014-08-15 Published:2014-08-04

摘要:

利用玛曲县尼玛镇研究区2009年8月30日的RapidEye影像和2011年8月27日的WorldView-2影像及同时段的MOD09GA产品,计算分析了基于高分影像的NDVI与基于MOD09GA的NDVI的相关关系,对基于MOD09GA NDVI的甘南州草地生物量模型进行了修正,并对修正后模型的估产精度进行了比较和评价。结果表明:基于MOD09GA NDVI(x)的甘南州草地生物量最优模型为指数模型(Y=1026.027e1.466x),其R2为0.571,高于其线性、乘幂、对数等模型;研究区内MOD09GA NDVI与依据RapidEye和WorldView-2计算的NDVI(分别记作NrNw)之间均具有显著的线性相关关系,且后者略优于前者,回归方程分别为Nr=1.2713x-0.2183和Nw=0.9559x+0.0679;结合MOD09GA NDVI与WorldView-2NDVI回归方程对甘南州MOD09GA NDVI与草地生物量模型进行修正后,得到的生物量更接近实测值,在研究区内平均估产精度可达84.73%,比未修正的估算模型精度提高1.08%。说明利用高分遥感影像修正的MODIS估产模型具有更好的精度。

关键词: MOD09GA, RapidEye, WorldView-2, NDVI, 相关性, 生物量

Abstract:

RapidEye image,WorldView-2 image and the simultaneous MOD09GA products of the study area in Nima Township were used to calculate and analyze the correlations between MOD09GA NDVI and NDVI of high-resolution satellite images in this paper. The regression model between MOD09GA NDVI and biomass measured in Gannan grassland was revised, and then the accuracies of the revised models were compared and assessed. The results indicate that, firstly, the optimal model between MOD09GA NDVI and the biomass in Gannan grassland is exponential model (Y = 1026.027e1.466x), the R2 value is higher than that of linear model, power model and logarithmic model. Secondly, MOD09GA NDVI has good correlations with both RapidEye NDVI (Nr) and WorldView-2 NDVI (Nw) in the study area, and the latter is superior to the former. The regression equations are Nr = 1.2713x-0.2183 and Nw = 0.9559x+0.0679. And then combining with the regression equation of MOD09GA NDVI and WorldView-2 NDVI, the biomass values simulated by the revised regression model between MOD09GA NDVI and biomass in Gannan grassland are much closer to the measured values, means that, the average yield estimation accuracy in the study area is 84.73%, which suggests that the revised model has a better accuracy and plays an important role in improving the simulation of grassland biomass in pastoral area.

Key words: MOD09GA, RapidEye, WorldView-2, NDVI, Correlation, Biomass

中图分类号: