The Correction ground spectral model for estimating above-ground net primary productivity at the peak of growing season on Meadow Steppe, Hulunbeier, Inner Mongolia, China

On the surface of hyperspectral data of remote sensing to estimate grassland biomass, the soil water content, organic matter content important impact on the model application. An ASD Fieldspec 3 spectroradiometer was used for spectral measurements of the Hulunbeier Meadow Steppe, Inner Mongolia, China, in late July, 2013. Ground spectral models were built to estimate the Above-ground Net Primary Productivity (ANPP) at the peak of the growing season from the Correction Normalized Difference Vegetation Index (C_NDVI) measured in the field. The SPSS software were used to assess relationships between ANPP and C_NDVI. Based on the coefficient of determination (R²), quadratic model (R²=0.808) and exponential model (R²=0.717) were batter than others. Error analysis shows that the linear equation has the biggest standard error of the prediction (SE = 82.42g/m²), the logarithmic curve equation has the smallest SE (SE=18.59g/m²). After considering all factors, a quadratic equation between ANPP and C_NDVI (ANPP=3738.048 NDVI_MODIS²-259.1588NDVI_MODIS+1309.847, R2=0.808, SE=66.25g/m², MEC=0.820, P <; 0.001) was selected and used for the study area at the peak of the growing season.