基于精确光谱响应匹配的星载成像光谱仪交叉辐射定标

Imaging Spectrometry Radiometric Cross-Calibration Based on Precise Spectral Response Matching

星载遥感器在轨运行中受到外太空环境以及遥感器自身特性衰变的影响,辐射特性会发生变化。为确保星载遥感数据能真实地反映被观测地物目标特征及其变化规律,需要定期对星载遥感器进行在轨辐射定标。环境小卫星超光谱成像仪(HJ1A/HSI)由于缺乏配套的星上定标系统,基于场地定标的方法难以满足高频次定标的需求。以EO-1/Hyperion为参考遥感器,以HJ1A/HSI为待定标遥感器,通过反卷积方法对两成像光谱仪光谱通道之间进行精确光谱响应匹配,消除波段设置的差异性,显著降低了HSI定标系数的不确定度。基于本定标方法得到的HSI 115个波段的绝对定标系数中,Band 1至Band 60之间的定标系数的不确定度稳定在5%～8%,除760nm附近的氧气吸收波段与940nm附近的水汽吸收波段外,其余波段的定标系数的不确定度为7%～18%,随着波长的增加,不确定度增大。与传统波段匹配方法相比,提高了约50%的精度,该定标精度基本可以满足遥感数据定量化应用的需求。该方法解决了在轨星载成像光谱仪光谱通道设置差异大、交叉定标精度低,难以实用的问题,为星载成像光谱仪高频率更新辐射定标数据提供了一种有效方法。

The present research describes the development of an improved cross-calibration method of on-orbit satellite sensor. The EO-1/Hyperion was taken as the referenced sensor and HJ-1A/HSI was taken as the uncalibrated sensor. The differences between the bands configurations were removed by the precise spectral response matching using the deconvolution method, which significantly reduced the radiometric calibration uncertainty of HSI sensor. The calibration coefficients of HSI for all 115 bands were acquired. The uncertainties of calibration coefficient from band 1 to band 60 stably lie in 5-8%, and for all the other bands excerpt for the oxygen absorption which lies in at 760 nra and the water vapor absorption which lies in at 940 nm, the uncertainties of calibration coefficients are changed from 7 % to 18%, which increased as the wavelength increased. Contrasted Compared with the traditional spectral matching method, the method proposed can improve the calibration accuracy by about 50%, which can meet the demand of the quantitive application for hyperspectral remote sensing data. It demonstrated the good precision and reliability of the method. It solved the spectral matching problem when the band configuration is big enough so that the cross calibration accuracy is too low and is difficult to apply in hyperspectral sensor cross-calibration, and provides a new method to frequently update the calibration coefficients for hyperspectral imager.