Research On The Influence Of Atmosphere's Status To Retrieval Of Land Surface Microwave Characteristics Using Satellite

The land surface plays an important role in the water, carbon, and energy cycles of the earth's climate system through land-atmosphere interaction. Therefore, it is important to accurately retrieve the land surface information and study the interaction between land surface and atmosphere.In this thesis,the retrieved ]and surface emission microwave dataset (including surface microwave brightness temperature and emissivity, etc.) by multi-source satellite remote sensing data were combined with other data (cloud product data, reanalysis data and precipitation data, etc.), to study two specific topics: 1. the influence of atmosphere to passive microwave retrieval of snow over Tibet Plateau; 2. the response of microwave land surface emissivity (MLSE) to precipitation in Amazon region.For the first topic, this thesis used the Advanced Microwave Scanning Radiometer-EOS (AMSR-E) upwelling brightness temperature on TOA (TBTIA) during 2002-2011 over the Tibet Plateau, and the upwelling brightness temperature at surface (TBSRF)which was derived based on microwave radiation transfer calculation by calibrating the effect of atmosphere and cloud. Using the two TBs, SDTOA and SDSRF were estimated,respectively. By comparing their difference and the relationship between the difference and cloud parameters and water vapor, the influence of atmosphere to passive microwave retrieval of snow was investigated.Through case analysis and nearly 10 years of statistics, it was found that: the atmosphere effect on TBs at low frequency microwave 18.7 GHz is weak while TBTOA at 37GHz are significantly warmer than TBSRF. Without considering such effect, the snow depth over Tibet Plateau would be underestimated (SDTOA<SDARF). The underestimations are common in multiple cases and significant at multi-year mean scales, therefore should not be neglected. Directly retrieving snow by TBTOA, The absolute error (SDTOA-SDARF) would be approximately 2-3cm. In the region with relative shallow snow, the relative error is up to 50-80%. While in the region with relative deep snow, the relative error is 10-20%. The error has strong negative correlation with liquid cloud water path (R=-0.45) with sensitivity of -0.047cm per g/m2.The error is not sensitive to ice cloud and even weaker to column water vapor. Snow extent retrieved from Moderate Resolution Imaging Spectroradiometer (MODIS) has better correlation to SDSRF than that to SDTOA. This imply that corrections of the influence of atmosphere can improve the accuracy of satellite passive microwave retrieval of snow depth over Tibet Plateau.In the second topic, by collocating the AMSR-E MLSE which was derived from multi-souce data and the Tropical Rainfall Measuring Mission/Precipitaion Radar(TRMM/PR) precipitation information data, the response of MLSE to precipitation occurred within 1 hour in Amazon region was investigated. The results show that this method of combining Aqua/AMSR-E and TRMM/PR observations is effective. The change of MLSE after rainfall is closely related to vegetation (or underlying surface).In dense vegetation areas, the average MLSE of samples in which rain was detected by PR (MLSEwet) did not differ significantly with that of all samples (MLSEAll), and even in some areas there are MLSEWet>MLSEAll. But in the areas with sparse vegetation,there are the relation of MLSEWet<MLSEAll. This result is likely caused by the interaction between precipitation and vegetation. The relation between MLSE and vegetation shows that with larger vegetation index, MLSE would be larger.Then, the relationship between MLSE and precipitation information (land surface rain rate and rain area) is investigated. The relation between surface rain rate and MLSE is negative, as well as rain area and MLSE. When the surface rain rate increased by lmm/hr, the MLSE would decrease by about 0.001-0.003. When the rain pixel ratio(which indicates rain area) increased by 0.1, the MLSE would also decrease by about 0.001-0.003. The two statistic results had passed the 95% confidence level test.The results of this study, on the one hand, would be able to provide the reference for developing more accurate algorithms of microwave remote sensing on land surface parameters (e.g. snow retrieval). On the other hand, it can be used to estimate MLSE when precipitation is happening in order to develop more advanced rain retrieval algorithm.