CORN FPAR ESTIMATING WITH NEAR AND SHORTWAVE INFRARED BANDS OF HYPERSPECTRAL DATA BASED ON PCA

摘要

FPAR (Fraction of Absorbed Photosynthetically Active Radiation) is a key parameter in the study on ecosystem function, crop growth monitoring, and so on, it is important to estimate FPAR accurately. Based on the analysis of measured corn hyperspectral and PAR data, the conclusions of this paper are: PCA approach can be used to distill hyperspectral information successfully, two principal components could hold more than 98.464% of the original hyperspectral information. PCA method could estimate FPAR effectively, for analyzing visible and near-infrared band with R2 of 0.858 and RMSE of 0.110, and analyzing near-infrared and shortwave band with R2 of 0.868 and RMSE of 0.106. Vegetation indices of (Normalized Difference Shortwave Index) and RSI (Ratio Shortwave Index), with the same structure with NDVI and RVI but calculated in different band, were better for FPAR estimation than NDVI and RVI. R2 for estimating FAPR of NDSI and RSI are 0.9026 and 0.8951, but 0.8510 and 0.8469 to NDVI and RVI. Near and shortwave hyperspectral reflectance has the great potential for estimating FPAR, which could be good to improve the precision of FPAR estimation.