有云条件下地表辐射量多传感器联合反演算法研究

As a main driving force for the earth's matter and energy cycle, the land surface radiation budget is a key modulator of global climate change. The corresponding surface radiative flux components (e.g., net radiative fluxes) are also the essential driving parameters for many models of land applications, ecosystem, hydrology and climate, etc. However, most existing retrieval algorithms of radiative fluxes are valid only under clear skies; thus, the corresponding products are highly discontinuous in the spatial scale when considering a globally averaged cloud cover of around 70%. Furthermore, to date few attempts at estimating radiative fluxes under cloudy skies are empirical or semi-empirically based, ignoring the actual spatial-temporal characteristics of clouds. The accuracies of the corresponding radiative flux products are therefore largely restricted. These shortfalls make current radiative flux products from space unsuitable for most land models and other applications. To overcome this challenge, we propose to develop a physically based fusion algorithm for key parameters that affect radiative flux estimations, such as profiles of temperatures and moistures, cloud physical and optical properties, and surface temperatures, etc., by synergizing currently available optical and microwave satellite data with an eye toward improving the accuracies and spatial continuities of those variables. Since different sources of data have different spatial and spectral properties, the cloud fraction, cloud distribution and point spread functions of the related sensors will be carefully considered during the data fusion. It should be noted that although sunlight may not be obstructed by clouds for certain pixels, the corresponding downward sky radiative fluxes (both shortwave and longwave) will likely be affected by hemispherical cloud cover and cloud positions which, unfortunately, cannot be modeled with current one-dimensional plane-parallel radiative transfer models. For rectifying this shortcoming, an algorithm for improving the downward sky radiative flux calculations is suggested by coupling the radiative transfer model to fully account for the hemispherical cloud fraction and cloud interactions over a specific spatial scale around the target. Based on this, both surface short- and long-wave radiative fluxes under cloudy skies will finally be retrieved by combining the required surface parameters such as, albedo and land surface temperatrue, etc. We believe that the proposed study will enhance links between application models and remote sensing, and will also provide a theoretical basis for studies of the earth's radiation and energy balances.

地表辐射收支是影响全球气候变化的关键要素，同时也是气候、气象、生态及水文等模型的重要驱动参数。现有辐射量遥感反演算法多数只适用于晴空条件，所得产品空间不连续；而有云条件下，光学遥感仅能反映云层上部或顶部信息，无法得到整个云层和云下大气、地表信息，只能依赖经验或半经验方法进行估算，反演精度有限，致使目前遥感反演的辐射产品与模型脱节，难于被广泛应用。本项目致力于联合微波(提供云层深部及云下信息)与光学(提供云覆盖、云分布等信息)遥感数据发展云参数、云下地表温度及大气廓线的融合算法，使各参量精度上达到最优。在此基础上，考虑目标像元与其半球空间云分布的关系(不再孤立的只针对像元本身)，耦合辐射传输模型发展下行辐射改进算法，进一步结合地表反照率、温度等实现有云条件下多个地表辐射分量的物理反演(短波及长波)，弥补当前有云条件下地表辐射量估算缺乏通用物理算法的不足，为地表辐射与能量平衡研究提供理论依据。

地表辐射收支是全球气候变化的关键影响要素和指示因子，同时也是气候、气象、生态及水文等模型的重要驱动参数。现有辐射量遥感反演算法多数只适用于晴空条件，所得产品空间不连续；而有云条件下，光学遥感仅能反映云层上部或顶部信息，无法得到整个云层和云下大气、地表信息， 只能依赖经验或半经验方法进行估算，反演精度有限，致使目前遥感反演的辐射产品与模型脱节，难于被广泛应用。.针对以上不足，本项目联合微波与光学遥感数据发展了云参数、云下地表温度及大气廓线的融合算法，通过多传感器数据的协同最终实现了有云条件下短波及长波辐射量的反演。由于计算有云条件下的地表辐射仍需要提前估算晴空条件下的辐射量，本研究利用MODTRAN模拟生成短波及长波辐射的模拟库，基于查找表的方法实现了晴空条件下短波及长波辐射的反演；同理，对于云下短波的估算，研究重新模拟得到模拟库，建立了针对有云条件的短波辐射查找表。由于云底高度、云底温度、云下地表温度及云下大气廓线等难以获得，也很难用模拟的方法估算，因此，研究采用多传感器融合的方法得到关键参量，最后由这些融合得到的关键参量计算得到有云条件下长波下行辐射和长波上行辐射量。.本研究基本建立了有云条件下短波及长波辐射量的反演技术体系，本研究的成果有效弥补了当前有云条件下地表辐射量估算缺乏通用物理算法的不足，切实能为地表辐射与能量平衡研究提供理论依据。.