Application of PCA and canopy near, shortwave-infrared bands for soybean and corn FPAR estimation in the Songnen Plain, China

Abstract

The fraction of photosynthetically active radiation (FPAR) absorbed by global vegetation is a key state variable in most ecosystem productivity models and in global models of climate, hydrology, biogeochemistry, and ecology. Therefore, how accurately retrieve FPAR will directly influence the estimation of many models and requires special attention. In this paper, based on the ground truth data in the Songnen Plain of China, we studied the correlations between FPAR and the corresponding vegetation indices. Comparing with NDVI and RVI (calculated by visible and near-infrared band), NDSI and RSI were also constructed (calculated by near, shortwave infrared band) to estimate FPAR. All vegetation indices were under the best wavelength combinations. PCA approach was also introduced for extracting hyperspectral reflectance information and estimating FPAR. The research results indicated that NDSI and RSI calculated by the near, shortwave infrared bands (R <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> of the validating models were 0.74 and 0.69 and RMSE were 0.108 and 0.171, respectively) showed better performance than NDVI and RVI that was computed by the visible and near-infrared bands (R <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> of the validating models were 0.71 and 0.65 and RMSE were 0.187 and 0.213, respectively). PCA approach could compress the hyperspectral reflectance information effectively, and showed better performance for FPAR estimating. From the above study, it also suggested that shortwave infrared bands had great potential for the estimation of FPAR.