【作者】 王雨；

【导师】 傅云飞；

【作者基本信息】 中国科学技术大学 ， 空间物理， 2008， 博士

【摘要】 随着科学技术的进步,越来越多的卫星（如TRMM）上搭载有多种高精度观测仪器,能够从各个角度实现对地气系统不同物理量的同步精确测量。基于这些仪器的综合观测结果,建立和发展多参数反演算法,有助于细致研究各种气候影响因素的变化及相互关系。本文利用TRMM上搭载的微波辐射计（TMI）、可见/红外扫描仪（VIRS）以及测雨雷达（PR）的同步观测结果,建立了一套可用于洋面非降水条件下的全参数（地表、水汽和多种云参数）同步反演算法—APRAMVI（All Parameters Retrieval Algorithm for combined Microwave andVisible/Infrared measurements）,并利用该算法,对中国近海区域的气溶胶间接气候效应进行了研究。1.APRAMVI算法APRAMVI算法由四个子反演算法组成,各子反演算法相互联系但技术上相对独立,分别为:（1）被动微波环境参数（风速、海温和水汽）反演算法:基于一个能够反映微波亮温和对其产生影响的五种参数（海温、风速、水汽、云水含量和云温）之间本质的对数线性关系的辐射传输方程,通过TMI五通道亮温对数线性组合,且无需其它任何辅助资料,实现对风速、海温和水汽的反演。作为算法的验证,地基观测数据和另一种基于TMI观测结果的反演产品被用于与本算法各环境参数反演结果的比较。结果表明,水汽反演值与探空资料的均值偏差(0.435 kg m^-2)和均方根误差(2.593 kg m^-2)都较小;与赤道地区浮标观测结果的比较表明,风速（海温）与实测值之间也仅有-0.075 m s^-1（0.116K）的均值偏差及0.672 m s^-1（0.665K）的均方根误差;而在全球尺度上,各反演参数与RSS TMI逐日产品的格点间（0.25°）差异统计特征也均表现为中值为0左右的准正态分布。并且,为研究其他参数对反演结果的影响程度,在验证过程中也按其他参数不同取值对各种反演参数进行了分类比较统计。结果表明,其它参数的变化,不会对反演结果造成太大影响,除了高风速时海温的反演结果有一定误差。此外,为了探讨算法在气候研究中的适用性,我们也对各反演参数的全球月平均分布情况进行了研究。结果表明,多年夏季（7月）和冬季（1月）反演参数的全球水平分布与实际情况基本相符,与其他卫星反演资料在全球大部分区域的偏差不超过1 m s^-1（风速）和1 K（海温）。（2）被动微波云参数（云水含量和云温）反演算法:利用对数线性关系辐射传输方程和环境参数反演算法的结果,对已有的一个云水含量反演框架进行了订正和改进,实现仅依赖于TMI观测结果的云水含量和云温同步反演。验证结果表明,有云时反演的云水含量与基于可见/红外算法获得的云水含量在分布上基本一致,均值偏差在0.03 kg m^-2左右（后者略高）;云温较海温低7K左右,也与实际大气相吻合;当无云时,仅有20%的无云像素被误识为云,并且平均误差仅为0.01kg m^-2;在全球尺度上的格点统计比较方面,反演结果要略低于RSSTMI产品的云水含量(0.007kg m^-2)。（3）综合被动微波和可见/红外的云参数（云厚和云高等）反演算法:利用被动微波环境参数反演算法和被动微波云参数反演算法的结果（海温和云温）,并结合VIRS对云顶温度的探测结果,基于一个简单的云模型实现对云厚和云高的反演。个例验证的结果显示,冰水混合云的云厚与云高均大于水云,定性表明了算法的合理性和准确性。（4）可见/红外云光学参数（有效半径和光学厚度）反演算法:利用VIRS可见光和近红外通道分别对光学厚度和有效半径敏感的性质,同步反演这两个参数（注:该可见/红外子反演算法技术源于已有文献,非本文成果）。2.中国近海气溶胶间接气候效应基于Terra和Aqua卫星MODIS气溶胶产品,对中国中东部地区气溶胶的一些物理特性,如气溶胶类型、光学厚度分布等进行了分析。研究结果表明,春夏季该区域气溶胶浓度较大,且在黄海沿岸地区,由于其位于季风下风向,更多来自中国南部的气溶胶粒子在此积累,导致该地区成为光学厚度极值区。相应地,其近海区域易出现较强的气溶胶间接效应。因此,利用APRAMVI算法的云参数反演结果（基于TRMM观测结果）,结合准同步的MODIS气溶胶产品,对春夏季中国东部近海区域的气溶胶间接效应进行了个例统计研究。结果表明,对于第二类间接效应,无论是单独的水云,或是全部的非降水云（包括冰水混合云）,均可被卫星直接观测到。但是,在中国近海这一特定区域,尽管有效半径与气溶胶光学厚度存在可通过信度检验（0.95）的负相关,但有效半径更多地是由云本身的热动力学过程决定,因此第一类间接效应虽真实存在,却无法被卫星直接观测到。更多还原

【Abstract】 Up to now, there are so many satellites having the capabilities to simultaneously observe various variables related to the Earth-Atmosphere System, with kinds of sensors in high-performance. Based on these united observations, it can be beneficial in the development and the improvement of multi-variables retrieval algorithm, in turn the investigation on the climate change associated to the variation of the variables and the relationship to each other. In this study, by using the combined measurements from TRMM TMI, VIRS, as well as PR, an integrated algorithm - APRAMVI （All Parameters Retrieval Algorithm for combined Microwave and Visible/Infrared measurements） is proposed to simultaneously retrieve surface and atmospheric variables including the sea surface temperature （SST）, wind （U）, columnar water vapor （CWV）, cloud liquid water path （LWP）, cloud temperature （Tc）, cloud geometric depth （Dc）, cloud height （Hc）, cloud effective droplet radius （Re）, cloudoptical thickness （τ_c）, and so on, over ocean in the absence of rain. Moreover, anapplication of this algorithm to explore the aerosol indirect effect over the Yellow Sea region is presented.1. APRAMVIThere are four sub-algorithms in APRAMVI, which are dependent on each other in retrieval step, but different in retrieval technique.（1） Environmental variables （SST, U and CWV） sub-algorithm for TMI measurements: Applying an inherent log-linear relationship between the microwave brightness temperature （T_B） and variables including CWV, SST, U, LWP and Tc, the TMI T_Bs on five channels are log-linearly combined to retrieve environmental variables. For the validation, the retrievals are compared to the ground-based measurements and the other TMI-based retrieval product, respectively. The results show that there are small differences between the retrievals and ground-based measurements, e.g., the mean bias （MB） and root mean square error （RMS） between CWV retrievals and radiosonde observations are 0.435 kg m^-2 and 2.593 kg m^-2, while MB and RMS between U （SST） retrievals and buoy observations are -0.075 m s^-1 （0.116K） and 0.672 m s^-1（0.665K）, respectively. On the other hand, a good agreement is also shown when global grid-to-grid （0.25°） comparing with the other TMI-based retrieval product （RSS TMI product）, where there is the mostly unbiased Gaussian probability distribution of the difference between them with maximum frequency near zero. Moreover, for various other variables, the variation of MB or RMS statistics is small, which suggests that the uncertainties related to other variables on any retrieved variable are eliminated in current sub-algorithm, except that there is significant error in retrieved SST with high wind speed. Finally, the results from monthly global distribution indicate the rationality of current sub-algorithm, due to low differences over mostly global ocean area with compared to other climatic dataset.（2） Cloud variables （LWP and Tc） sub-algorithm for TMI measurements: Based on the above log-linear relationship and environmental retrievals, a published LWP retrieval frame is corrected and improved to simultaneously retrieve LWP_m （subscript m indicates LWP obtained from TMI measurements）and Tc with only TMI measurements as the input data. The quantitative validation shows that, with respect to cloud presence, the retrieved LWP_m is consistent to the retrieved LWP_s （subscript m indicates LWP obtained from VIRS measurements） by Visible/Infrared sub-algorithm （See subsequent paragraph about it）, whether the pattern or values. Specially, the MB about 0.03 kg m^-2 （LWP_m is low） is in agreement with the generally intrinsic difference between them; the difference about 7 K between retrieved Tc and SST is coherent with actual atmosphere. For cloud absence, only 20% of clear-sky pixels are misjudged as cloud pixels by current sub-algorithm, where the average value is less than 0.01 kg m^-2; the grid-to-grid comparison with RSS TMI product on global scale indicates the slight difference of 0.007 kg m^-2 between them.（3） Cloud variables （Dc and Hc） sub-algorithm for combined TMI and VIRS measurements: Based on a simple cloud model, the conjunction of above retrieved SST and Tc for TMI measurements and cloud top temperature detected by VIRS is used to retrieve Dc and Hc. Case study results show that Dc and Hc of multi-layer ice-water mixed cloud are larger than single-layer water cloud, which qualitatively suggests the rationality and accuracy of current sub-algorithm.（4） Cloud optical variables （Re,τ_c and LWP_s） sub-algorithm for VIRS measurements: Based on the separate dependence of VIRS visible and near-infrared channel onτ_c and Re, combination of these two channels is used to simultaneously retrieveτ_c and Re. The retrieval technology for this sub-algorithm is derived from published references.2. The aerosol indirect effect over the Yellow Sea regionFirst, the aerosol characteristics in middle-eastern China, such as aerosol optical depth, aerosol type, mass concentration and fine aerosol fraction, are investigated by using the MODIS aerosol product from Terra and Aqua satellites. The results indicate that there is high aerosol concentration in this region during Spring and Summer. Specially, the largest aerosol optical depth locates the region near the Yellow Sea, due to aerosol accumulation effect with the prevailing west-south wind direction in East Asia summer monsoon period. Similarly, it is possible that there is obvious aerosol indirect effect over the Yellow Sea. As a result, the aerosol-cloud relationship is studied by using retrieved cloud parameters derived from APRAMVI for TRMM measurements and aerosol optical depth in MODIS product. The cases statistics results show that, the second aerosol indirect effect is obvious over the Yellow Sea region, whether for water clouds or for all non-precipitation clouds. However, the first aerosol indirect effect can not be observed by satellite, though there is significant negative correlation between Re and aerosol optical depth. Further investigation on the relationship between Re and cloud thermodynamic parameters suggests that the influence of thermodynamics on Re exceeds that of aerosol, so that the first aerosol indirect effect is obscure in this particular region.更多还原