Development of global estimation algorithms for soil moisture and vegetation amount using Tropical Rainfall Measuring Mission (TRMM) data

本研究利用热带降雨测量任务（TRMM）上的两个微波传感器——降水雷达（PR）和TRMM微波成像仪（TMI）制作全球地表土壤湿度图。首先，提出了一种基于TRMM/PR的土壤水分估算算法。该算法采用TRMM/PR观测到的地表后向散射系数。由于强降雨会减弱后向散射系数，因此降雨期间的观测数据不包括在我们的计算中。与合成孔径雷达（SAR）相比，TRMM/PR的空间分辨率较差，但观测频率（时间分辨率）与无源微波传感器相当。虽然TRMM/PR采用0 ~ 18度多个入射角观测，但我们的算法基本上是针对12度观测设计的。灵敏度分析表明，在12度观测时，后向散射系数受地表植被覆盖度的影响较小。然而，当植被覆盖度的时间变化不显著时，3 ~ 18度的观测值与12度的观测值具有较好的相关性。在这种情况下，我们的算法可以应用于3到18度的大量观测。其次，提出了一种基于TRMM/TMI的土壤水分估算算法。利用土壤-植被-大气层微波传输模型，从亮度温度的偏振差/频差等单指标反演土壤水分。对不同算法估算的土壤湿度进行了比较，并与TRMM/PR估算的土壤湿度进行了比较。为了模拟真实的土壤水分季节变化，有必要提供其他卫星信息的叶面积指数（LAI）数据。其他植被参数(如：茎面积指数（SAI）和土壤粗糙度也很重要，因为它们显著影响PD/FD对LAI的敏感性。10GHz的PD受大气影响较小，对土壤湿度具有较高的敏感性，是反演土壤水分的最佳指标。最后，将这两种算法分别应用于1998-2000年3年全球地表土壤湿度日步长数据集。研究了这些估计的日变化，并讨论了两种方法的优点。少

Two microwave sensors on board Tropical Rainfall Measuring Mission (TRMM), Precipitation Radar (PR) and TRMM Microwave Imager (TMI), are used to make gtobalsurface soil moisture map in this research. Firstly, a soil moisture estimation algorithm from TRMM/PR is developed. In this algorithm, the backscattering coefficients at land surface observed by TRMM/PR are used. As the backscattering coefficients are attenuated by strong rainfall, the data observed during rainfall is not included in our calculation. TRMM/PR has poor spatial resolution compared with Synthetic Aperture Radar (SAR), but the observation frequency (temporal resolution) is as high as passive microwave sensors. Though TRMM/PR observes by multiple incident angles from 0 to 18 degree, our algorithm is basically designed for the observations by 12 degree. It is shown by a sensitivity analysis that the backscattering coefficients are less affected by the ratio of surface vegetation cover when they are observed by 12 degree. … More However, if the temporal change of vegetation cover ratio is not significant, the observations by among 3 to 18 degree are well correlated with those by 12 degree. In such case, our algorithm can be applied to large number of observations by 3 to 18 degrees. Secondly, another soil moisture estimation algorithm from TRMM/TMI is developed. A microwave transfer model in soil-vegetation-atmosphere layers is used to retrieve soil moisture from single index such as polarization difference (PD) / frequency difference (FD) of brightness temperatures. Soil moisture estimates from different algorithms are compared to each other and are compared with estimates from TRMM/PR. In order to simulate realistic seasonal variation of soil moisture, it is necessary to give monthly Leaf Area Index (LAI) data from other satellite information. Other vegetation parameters (eg. Stem Area Index (SAI)) and soil roughness are also important because they significantly affect the sensitivity of PD/FD against LAI. PD of 10GHz is the best index to retrieve soil moisture because this is less affected by atmosphere and this has high sensitivily against soil moisture. Finally, these two algorithms by two sensors are applied to make global dataset of surface soil moisture in three years (1998-2000) with daily time step. Diurnal variations of these estimates are examined and the advantages of both methods are discussed. Less