资源一号02D高光谱数据大气校正及应用

Atmospheric Correction and Application of ZY-1-02D Hyperspectral Data

针对高光谱数据大气校正耗时长和查找表构建不准确等问题,提出基于MODTRAN辐射传输模型实时创建大气校正参数查找表的方法,并应用于水体叶绿素浓度反演。首先,基于高光谱数据实时构建大气校正参数查找表;其次,根据循环迭代反演得到水汽含量和气溶胶光学厚度对查找表插值得到各个波段的大气校正参数,从而完成所有波段数据的大气校正;最后,选择植被、土壤和水体3类典型地物精度分析,并基于反演水体的叶绿素a浓度验证大气校正精度的可靠性。实验结果表明:该方法明显优于6S、FLAASH等大气校正方法;在运行效率上,在多线程并行加速后,运行效率提升了2~4倍;基于水体反射率数据反演水体叶绿素a浓度,反演模型预测集验证中ρ为0.804 7,RMSE为1.8。

Aiming at the problems of time-consuming atmospheric correction of hyperspectral data and inaccurate lookup table construction,a method of real-time creation of atmospheric correction parameter lookup table based on MODTRAN radiation transmission model is proposed and applied to water body chlorophyll concentration inversion.Firstly,build a lookup table of atmospheric correction parameters in real time based on hyperspectral data.Secondly,interpolate the lookup table based on the iterative inversion of loops to obtain the water vapor content and aerosol optical thickness to interpolate the lookup table to obtain the atmospheric correction parameters for each band,thereby completing the atmospheric correction of all band data.Finally,three types of typical features of vegetation,soil and water are selected for accuracy analysis,and the reliability of the atmospheric correction accuracy is verified based on the inversion of the chlorophyll a concentration of the water.Experimental results show that this method is significantly better than 6 S,FLAASH and other atmospheric correction methods.In terms of operating efficiency,after the algorithm is accelerated in parallel by multiple threads,the operating efficiency is increased by two to four times.The chlorophyll a concentration of the water body is inverted based on the water body reflectivity data.In the verification of the inversion model prediction set,ρ is 0.804 7,and the RMSE is 1.8.