若尔盖白河牧场地上生物量与遥感植被指数关系研究

摘要

To study the aboveground biomass using remote sensing (RS) on the grassland, and serve better for the protection area, the TM images technology was used to research the relationship between aboveground biomass and RS Vegetation Indexes of Baihe Pasture of Zoige County. The linear and 6 nonlinear (Logarithmic, Inverse functions, Quadratic polynomial, Cubic polynomial, Composite and Power function) regression models were established respectively to estimate the pertinence between 7 vegetation indexes (NDVI, RVI, DVI, SAVI, MSAVI, PVI, GVI) and aboveground biomass. Results indicated that, the cubic polynomial regression model was the best model for the 7 vegetation indexes, followed by the quadratic polynomial model, inverse function curve model, logarithmic curve model, linear model, power function curve model and composite curve model. The cubic polynomial model based on MSAVI -aboveground biomass was the best model, the multiple correlation coefficient R2, average error, rating precision was 0.823005, 38.7%, 61.3% respectively. This model was able to meet the mesoscale estimates of the aboveground biomass.%目的:为探讨利用遥感影像进行草原地上生物量估测,更好服务于保护区内的植被生态建设.方法:本文采用TM影像研究若尔盖地区白河牧场地上生物量与遥感植被指数的关系,分别建立了7种植被指数( NDVI、RVI、DVI、SAVI、MSAVI、PVI、GVI)与地上生物量的线性和6种非线性(对数、反函数、二次多项式、三次多项式、复合、幂函数)回归模型.结果:植被指数MSAVI、NDVI、SAVI、RVI、GVI、DVI、PVI与白河牧场地上生物量模型表现出三次多项式回归模型最优,其次是二次多项式、反函数曲线模型、对数曲线模型、线性模型、幂函数曲线模型、复合曲线模型.结论:基于MSAVI的地上生物量的三次多项式模型的模拟效果最好,复相关系数R2=0.823005,精度检验结果表明该模型的平均误差为38.7％,拟合精度达到61.3％,能够满足中尺度地上生物量的估测.