Evaluation of Sensor Spectral Parameters for the Simulation Accuracy of the Vegetation Spectrum
LI Bo1,2，YAN Lei1,2*，ZHANG Li-fu3
1. Institute of Remote Sensing & GIS, Peking University, Beijing 100871, China 2. Beijing Key Lab of Spatial Information Integration and 3S Engineering Applications, Peking University, Beijing 100871, China 3. Institute of Remote Sensing Applications, Chinese Academy of Sciences, Beijing 100101, China
Abstract：Hyperspectral imaging (HSI) has become one of the most promising and emerging techniques in remote sensing. Due to hundreds of co-registered bands used in HSI system, hyperspectral imagery may provide more spectral information than multi-band images. Unfortunately, original hyperspectral images are more expensive and difficult to achieve than multi-band ones. However, an abundance of spectral information has to be acquired by part of special research for the purpose of ground monitoring, which original HSI systems can easily provide. Then a solution, called hyperspectral satellite data simulation, is proposed for studies in satellite data simulation. It is also one of the most important studies to simulate satellite remote sensing data. In the method, the model with low computational complexity can simulate hyperspectral data quickly, which is based on the priori spectral knowledge of the ground objects. But the accuracy of the simulation data depends on spectral parameters of the sensor. In the present paper, the authors experiment with EO-1/ALI bands in VIS/NIR wavelengths. Then the relationship between the spectral parameters, including the number of bands, bandwidth and the peak wavelength, and the simulation accuracy of the vegetation spectrum are analyzed from their variation principles. According to the results, spectral parameters can determine the effective spectral feature of the vegetation, and impact simulation model directly. Optimal parameters are also summarized for spectral reconstruction in the paper. The experiment results are beneficial to enhancing spectral simulation precision. The conclusions can help evaluate the performance of multispectral sensors and perfect spectroscope and filter design.
