Satellite Remote Sensing And Numerical Modeling Of Microwave Land Surface Radiative Characteristics Under All-weather Conditions

At present,passive microwave remote sensing is widely applied,beacasue of its sensitivity to water abundance,being less affected by the atmosphere and clouds,and the effective penetrating detection into the vegetation canopy and soil,in the retrieval and monitoring of atmospheric vapor,cloud water,precipitation,soil moisture and vegetation moisture content,snow cover,sea ice,and other hydrological-related research fields.Microwave Land Surface Emissivity(MLSE),termed as a intrinsic surface microwave radiative parameter,has important application values in atmospheric parameter retrieval,monitoring the dynamics of land-vegetation-atmosphere water and carbon cycling,and numerical weather forecast(NWP).This paper is mainly dedicated to two key related questions,On the one hand,although various surface emissivity acquisition algorithms have basically satisfied the demends of scientific research and operational of NWP,most of the existing work is carried out under the condition of clear sky,and the research/algorithm for instantaneous MLSE retrieving under cloudy condition of cloud is scarcely exploited.Two major problems appear in this circumstance:cloud cover leading to substanital gaps in satellite retrievals;the truth surface hydrological dynamics under cloudy skies cannot be detected.In order to achieve instantaneous microwave surface emissivity retrieving in the presence of clouds,a general algorithm framework for variational inversion is proposed in this study,which combines microwave observations of spaceborne microwave imager,cloud parameters retrievals of synchronous visible and infrared spectrometers,and high-resolution atmospheric&surface parameter reanalysis data.The general algorithm is applied to the retrievals of MLSE using AMSR-E,GMI and Fengyun 3B MWRI observations,respectively.On the other hand,accurate simulation of surface emissivity based on physical principles is the key to accurately retrieve surface hydrological parameters such as soil moisture and vegetation water content using spaceborne observations.However,the accuracy of surface parameterization and the inadequate understanding of its microwave characteristics restrict the progress of surface microwave radiation simulation.As an extension of satellite MLSE retrieval,this paper studied the radiation characteristics of major soil and vegetation parameters by modeling surface microwave radiation transmission in bare soil and vegetation scenarios,and supported the response feature of Emissivity Difference Vegetation Index(EDVI)found by satellite retrieval to vegetation water content and its vertical structure in forest canopy.The main conclusions of this paper are as follows:(1)The evaluations of AMSR-E,GMI and FY3B MWRI all-weather MLSE retrievals show that the general retrieving algorithm and the corresponding retrievals proposed in this study have achieved highly consistent performances with similar products in the world,indicating that our algorithm is reasonable and feasible,and the data quality is reliable.The corresponding correlation coefficient(R)to the existing products for AMSR-E MLSE was within 0.71-0.95,and the mean bias(MB)was less than 0.02;For Fengyun3b MWRI MLSE,R was within 0.87-0.92,and the MB value was within only-0.001-0.024;For GMI MLSE,R ranged from 0.835 to 0.978,and the MB value was less than 0.02(less than 0.002 in forest area).In terms of error sources,the error of land surface temperature and zenith brightness temperature(version)is the largest,with the difference of 3 K causing about 1%error in MLSE.In addition,every 300g/m2 error of cloud liquid water path contributes about 1%error to MLSE retrieval.(2)The surface emissivity obtained by series retrievings has obvious spatial heterogeneity:the MLSE of forest area presents a moderate to higher value and a smaller polarization difference;Lower values are found in river basins,coastal areas and farmlands.For desert and bare soil,the vertical polarization and polarization difference are at much higher level.In terms of seasonal variation,MLSE was highly correlated with the seasonality of vegetation growth,rainfall and snow cover.(3)The emissivity retrieving algorithm by combining Himawari-8 AHI cloud properties and Fengyun 3B MWRI observations is feasible.The data set has reached the accuracy level of similar international products,and the trend of annual change is stable.(4)Based on the study of diurnal variation characteristics of GMI MLSE,it was found that the MLSE of Borneo and New Guinea rainforests continued to rise during 7:00 am to 10:00 am,and then gradually decreased during 10:00 am to 14:00 am.In the sparse vegetation areas such as AUS and Huabei,MLSE showed a continuous decrease to the minimum value from morning to 14:00 PM and then followed with a slow increase.Based on the correlation analysis and model simulation analysis,we found that the unique MLSE diurnal variation patterns in the rainforest and sparse grassland areas were mainly attributed to the diurnal variation differences in the consumingrefilling-banlance of the vegetation water content under different soil moisture and evapotranspiration conditions.(5)The study of microwave surface emissivity characteristics based on surface radiative tranfer modelings showed that MLSE of bare soil decreased with the increase of soil moisture.For dry soil,the higher the vegetation water content,the lower the vertical polarization emissivity.When soil moisture is high,the surface emissivity first increases and then decreases with the increase of leaf moisture content.When the leaf moisture content remained unchanged,MLSE increased with the increase of Leaf Area Index(LAI)and eventually tended to a constant value.The MLSE based on surface simulation and satellite inversion achieved a good agreement in the forest area,but there was a big difference in the sparse vegetation area.(6)Based on satellite retrieving and model simulation,it is confirmed that multichannel EDVIs have different ability to represent the vegetation water content at different canopy levels:The low-frequency EDVI index has stronger penetration into the forest canopy thus can represent more levels of water content,resulting in EDVIc and EDVIX being significantly larger than EDVIKu,especially during the growing season when vegetation water content is the highest.High frequency EDVI was less suppressed by soil moisture in sparse vegetation areas,resulting in EDVIc and EDVIx lower than EDVIKu.With the increase of forest height,EDVIc(EDVIx)become gradually larger than EDVIX(EDVIKu),which is because the canopy gradually shields the suppression effect of soil moisture.