【作者】 张连义；

【导师】 刘德福；

【作者基本信息】 内蒙古农业大学 ， 草业科学， 2006， 博士

【摘要】 以锡林郭勒盟为研究区域,将遥感技术用于草地牧草产量测量,分析了遥感植被指数与草地牧草产量之间的相关关系,比较和分析了三种植被指数的应用范围,研究了草地牧草产量遥感监测的方法,并利用遥感植被指数建立了草地牧草产量估产模型,为今后开展大面积草地牧草产量估产和动态监测提供了有效途径。本文的主要研究结论如下:1.通过三种植被指数与草地总牧草产量的线性回归分析和非线性回归分析,可以得出:锡林郭勒草地总牧草产量估产的植被指数是NDVI,其估产模型是S曲线回归模型。2 .通过对锡林郭勒盟适宜的景观生态区域划分,可以得出:草甸草原区的估产模型是一元线性回归模型;典型草原区的估产模型是一元线性回归模型;荒漠草原区的估产模型是S曲线回归模型;沙地植被区的估产模型是幂函数回归模型。3.锡林郭勒盟草甸草原区、典型草原区、荒漠草原区和沙地植被区实测草地牧草产量与三种不同植被指数的相关性存在一定差别,其中EVI与草甸草原区的相关性最好;NDVI与典型草原区和沙地植被区的相关性最好;SAVI与荒漠草原区的相关性最好。4.通过锡林郭勒盟2001年和2005年实测点的预测值与实测值的误差统计分析:2001年锡林郭勒草原区总牧草产量的回归分析,估产精度达到91.6%;2005年锡林郭勒草原区总牧草产量的回归分析,估产精度达到87.1%。但是通过对锡林郭勒盟适宜的估产区域划分,估产精度得到进一步提高,其中草甸草原区一元线性回归模型的估产精度达到97%;典型草原区一元线性回归模型的估产精度达到97%;荒漠草原区S曲线回归模型的估产精度达到97%;沙地植被区幂函数曲线回归模型的估产精度达到98%。5. MODIS/NDVI植被指数曲线和估产牧草产量曲线能很好地反映草地类型的生长季节变化过程,这表明植被指数与草地牧草产量之间有非常显著的相关性,同时也说明锡林郭勒盟草地牧草产量总估产模型能够满足实际估产的需要。更多还原

【Abstract】 The remote sensing technique was used to measure biomass in Xilinguole League. The correlation between the vegetation indics and the biomass was analysed. Application range for three types of vegetation indices and biomass monitoring method by remote sensing was discussed. Models based on vegetation index were set up for estimating biomass and dynamically vegetation monitoring in large area. Main results and conclusions from the research are as follows:1. Linear and nonlinear regression relationships show as follows: NDVI has the best correlation to the region of Xilinguole;S regression model is suitable for the region of Xilinguole.2. It can be concluded from the suitable landscape ecological regionalization that:Linear Regression Model are suitable for both meadow steppeand typical steppe;S Regression Model is suitable for desert steppe;Growth Regression Model is suitable for sand grassland.3. The grassland biomasses of meadow steppe, typical steppe, desert steppe and sand grassland have different correlations with three types of vegetation indices,EVI has the best correlation with meadow steppe;NDVI has the best correlation with typical steppe and sand grassland;SAVI has the best correlation with desert steppe。4. By the analysis of estimating errors in 2001 and 2005,the results show that the estimating accuracy of S regression model is 91.6% to the region of Xilinguole in 2001 and that is 87.1% in 2005。Estimating accuracy for the grassland biomass is improved further by dividing the different region for estimating biomass. The estimating accuracy of linear regression model is 97% to meadow steppe. The estimating accuracy of linear regression model is 97% to typical steppe. The estimating accuracy of S regression model is 97% to desert steppe. The estimating accuracy of growth regression model is 98% to sand grassland。5 Vegetation indics and estimating biomass’curve can well reflect the change of the four types of grasslands. This shows that relationship between NDVI and biomass production can be expained well,and the general regression model can meet the need of estimating biomass in Xilinguole。更多还原