基于FY-3C微波成像仪的沙漠地表发射率反演

Retrieval of desert land surface emissivity based on FY-3C microwave radiation imager

地表发射率是地表的固有属性,也是反演地表信息和大气温湿度廓线的重要参数.为了获取准确且具有具体物理含义的沙漠地区微波地表发射率,首先选取塔克拉玛干沙漠部分地区为反演区域,根据二元函数泰勒定理,推导了该地区的微波地表发射率与地表温度、地表湿度的线性、非线性函数关系.其次,利用最优控制原理,结合FY-3C微波成像仪的观测亮温资料与辐射传输模式(CRTM)模拟亮温数据,构建了沙漠地区微波地表发射率的线性与非线性反演模型.通过对比发现,利用线性和非线性反演模型得到的地表发射率不仅提高了反演区域亮温的模拟精度,而且模拟亮温的变化趋势也与观测更吻合.最后,对地表发射率的线性和非线性反演模型进行了不同时间与空间上的独立性检验,结果表明:除了反演区域外,在整个塔克拉玛干沙漠地区,两种模型反演的地表发射率仍比原地表发射率模拟亮温更接近观测.总的来说,线性和非线性反演模型对沙漠地区的微波地表发射率反演均具有一定的有效性和普适性,且非线性反演模型优于线性反演模型.

Land surface emissivity is an inherent property of land surface,an important condition for retrieving surface and atmospheric parameters,and a key element in land surface data assimilation system.In order to obtain more accurate and physically meaningful microwave land surface emissivity,Firstly,this paper takes Taklimakan desert as the study area,and derives the linear and nonlinear functions of microwave land surface emissivity with respect to surface temperature and surface humidity by Taylor expansion.Secondly,according to the principle of atmospheric radiative transimission and the theory of optimal control,two cost functions are established by combining the observed brightness temperature with its simulated counterpart.Then,the linear and nonlinear retrieval models of microwave land surface emissivity in the hinterland of the Taklimakan desert are obtained by minimizing the cost functions with the Newton iteration method.The results show that the two cost functions decrease quickly,the simulated brightness temperature of CRTM model is very sensitive to the change of land surface emissivity,and the simulated brightness temperature with the retrieval land surface emissivity(new)is more consistent with the observed brightness temperature than the simulated brightness temperature calculated by the original land surface emissivity(old).Furthermore,the spatial and temporal independent tests of the linear and nonlinear retrieval models of land surface emissivity are carried out.It is found that the simulated brightness temperatures calculated by the two retrieval models are obviously improved compared with the old simulated brightness temperatures in the whole Taklimakan desert.The two retrieval models in this paper are also applicable to the retrieval of microwave land surface emissivity for the other desert areas.In general,the nonlinear retrieval model is better than the linear retrieval model.