在轨高光谱传感器光谱性能参数的准确定标是数据定量应用的基本前提。 文章在前人基础上, 综合优化算法, 实现了在不需要实测地表反射率的情况下, 同时反演高光谱传感器中心波长与半值波宽(full width at half maximum, FWHM)。 基于模拟数据的研究结果显示, 该方法在光谱性能参数偏移5 nm时, 中心波长反演误差小于0.1 nm, FWHM误差小于0.7 nm。 将该方法应用于Hyperion数据, 结果显示, Hyperion在VNIR谱段存在明显的smile效应, 在整个CCD阵列范围内, 其中心波长的偏移量在-2～2 nm之间, FWHM偏移在-0.2～0.5 nm之间； 在SWIR谱段smile效应不明显, 其中心波长偏移3 nm左右, FWHM偏移在-2～-3 nm之间。 最后在光谱重定标基础上, 对Hyperion进行了大气校正, 反演了不受大气及定标参数影响的地表反射率。 经光谱重定标, 在大气吸收波段周围由光谱定标参数变化导致的反射率突变得到了抑制.

On-orbit spectral calibration of hyperspectral imaging data is a key step for quantitatively analyzing them. Like the atmospheric correction, accurate spectral calibration is very necessary for improved studies of land or ocean surface properties. Based on the previous literatures, a new method which coupled an optimization algorithm was developed to simultaneously retrieve the central wavelength and the full width at half maximum (FWHM) of the hyperspectral sensor without needing the in situ reflectance spectra. Firstly, the Hyperion data set simulated using MODTRAN4 with the Hyperion spectral specification was used to test the new method, and the results indicated that the maximum error was less than 0.1 and 0.7 nm for central wavelength and FWHM respectively when the spectral shift is 5 nm. Then the algorithm was applied to the Hyperion data acquired on May 20, 2008 over Heihe River Basin and it was iteratively performed for each detector of the two spectrometers of Hyperion. The results showed that the VNIR of Hyperion had a pronounced smile effect, and the shift in on-orbit calibration with respect to the laboratory was from -2 to +2 nm, while the SWIR has essentially no smile effect, the wavelength correction was relatively flat over all sample with an approximately constant value of 3 nm. The FWHM in VNIR could range from -0.2 to 0.5 nm as a function of sample number of the spectrometer, and in SWIR it ranged from -2 to -3 nm. So for both the VNIR and SWIR, the original spectral calibration should be updated. These results showed good agreement with previous research findings, and which also proved the feasibility of the new method. Finally, with the updated spectral calibration characteristics, the sample reflectances of desert and vegetation target in our study site were reconstructed by applying a further atmospheric correction, and as expected, the strong spikes around the typical atmospheric absorption were almost disappeared.