沙漠地区土壤质地对不同频点微波地表发射率反演和模拟的影响

Effects of soil texture on the retrieved microwave emissivity at the different frequencies of a desert area and its modeling

中文关键词:  先进微波扫描辐射计（AMSR-E）  土壤质地  地表微波发射率模型  沙漠地区

英文关键词:Advanced Microwave Scanning Radiometer-Earth Observation System (AMSR-E)  Soil texture  Microwave land emissivity model  Desert area

中文摘要:

      首先运用先进微波扫描辐射仪（AMSR-E）资料反演了北非沙漠地区晴空条件下的地表微波发射率。然后根据不同的土壤类型，进一步分析了沙漠地表微波发射率频谱特性，并将增加土壤质地信息前、后的翁（Weng）氏微波地表发射率模型（2001）的模拟结果和反演结果进行了比较。结果表明，沙漠地表发射率与土壤质地密切相关，随土壤颗粒大小的不同变化明显。在沙漠土壤类型中，以大颗粒为主的土壤类型，其水平（垂直）极化的发射率通常随频率提高而增大（减小）；而对于以较小粒子为主的沙漠土壤类型，地表发射率几乎为常数，或水平（垂直）极化的发射率随频率提高略有减小（增大）。并且，发射率的季节性特征明显，特别是以小颗粒组成的土壤，其水平极化的发射率比垂直极化的发射率表现出更强的季节性变化。以上这些发射率特征与翁氏模型模拟结果一致。此外，在翁氏模型的输入参数中增加土壤质地信息（土壤组分含量、粒径尺度）改善了翁氏模型在沙漠地区的模拟结果，特别是对于包含大量小粒子的沙漠土壤类型，如黏土和黏质壤土，模拟误差从6%-9%降低至4%以下。由于翁氏模型是美国国家环境预报中心（NCEP）全球同化和预报系统的重要组成部分，对翁氏模型的改进将提高沙漠地区卫星资料的利用率并有望改进数值天气预报的准确度。

英文摘要:

      In this study, the surface microwave emissivity of North Africa deserts is first retrieved from the AMSR-E (the Advanced Microwave Scanning Radiometer-Earth Observing System) measurements under clear atmospheric conditions. Then, the spectral characteristics of the retrieved emissivity are classified and analyzed with respect to the underlying soil texture types. The analyses indicate that the desert microwave emissivity is closely related with soil texture, and significantly varies with soil particle size and composition. For those soil texture types which are mainly dominated by large-size particles, the emissivity spectra of vertical (horizontal) polarization generally decrease (increase) with frequency. Yet, for those soil texture types which contain fairly small-size particles, the emissivity values are almost constant or slightly increasing (decreasing) with frequency for the vertical (horizontal) polarization. Furthermore, the microwave emissivity of the desert area also exhibits some seasonal variation, especially for the soils composed by small-size particles. The horizontal polarization (H-Pol) emissivity shows stronger seasonal variability than the vertical polarization (V-Pol) emissivity. These desert emissivity features are generally consistent with the simulations by the microwave land emissivity model of Weng et al (2001). The performance of Weng's model over the desert area has been greatly improved with its input parameters explicitly taking into account soil texture information including sand fraction and particle size, e.g., emissivity simulation bias of clay and clay loam decreased from 6%,even 9% to less than 4%. Since Weng's model is an essential component in the National Centers for Environmental Prediction (NCEP) Global Forecast System (GFS), it is expected that such improvement would enhance the data assimilation over the desert area, and subsequently improve the accuracy of the numerical weather prediction (NWP) system.