**Surface Upwelling Longwave Radiation Estimation Model and Methodology Study Considering Thermal Directionality**

Author:Hu Tian

Supervisor:Willow admire fire, du Yongming

Database:Doctor

Degree Year:2018

Download:83

Pages:121Size:6690K

Keyword:Combined Algorithm，Direct Physical Algorithm，Directional Emissivity Look-up Table，Surface Upwelling Longwave Radiation (SULR)

Surface upwelling longwave radiation(SULR)is an index that reflects the thermal condition of the land surface,which consists of the emittance from the land surface from 4 to 100μm and the reflected atmospheric downwelling longwave radiation.SULR is an important component of surface radiation budget(SRB)and closely related to the temperature rise or cooling of the land surface and energy consuming in the evapotranspiration of the land surface and low layer atmosphere,thereby directly affecting the thermal and humidity environment that human beings and other creatures live in.During the night and most time of the year in the polar regions,SRB is dominated by SULR.Thus,accurately estimating the SULR is essential to the study of the surface energy circulation and climate change.Remote sensing can provide the access to regional and global observation of the Earth on a sufficient temporal and spatial resolution at a relatively low cost.Thus,remote sensing becomes the only choice to estimate the SULR at a regional or even global scale.Currently,there are mainly two types of methods to estimate the SULR using remote sensing:the first one is the physical temperature-emissivity algorithm;the other is the hybrid method using the top-of-atmosphere(TOA)radiance.In both categories,the land surface is assumed to be Lambertian.And the thermal directionality is not considered in the estimation methods,which may induce large errors in the estimation,particularly over sparsely vegetated surfaces.Therefore,this dissertation is aimed at considering the thermal directionality in the estimation of SULR and proposing different methods of estimating the SULR based on the kernel-driven model.The proposed methods were validated using the airborne data(WiDAS)collected during the Heihe ecological and hydrological comprehensive experiment(HiWATER).Moreover,at the satellite scale,a look-up table(LUT)of the directional emissivity for different land surface types in different seasons was built based on the MYD21 land surface temperature(LST)and emissivity product.And SULR was estimated using the LST from the MYD21 product and the hemispheric emissivity calculated from the LUT.The estimation was validated using measurements from the SURFRAD network over the CONUS.The effect of thermal directionality on the SULR estimation is very notable over sparsely vegetated surfaces.Thus,this thesis starts from the research on vegetated surfaces.The effect of thermal directionality on the SULR at different LAI cases was analyzed using the 4SAIL model first.Based on the analysis,the combined algorithm was proposed.In this algorithm,the FRA97 model and the kernel-driven model were used to consider the directional emissivity and the directional thermal radiance,respectively.And these two models were coupled in the combined algorithm using the traditional temperature-emissivity algorithm.The estimation from the combined algorithm was validated using the measurements from automatic weather stations(AWS)in the HiWATER experiment.The results showed that the estimation accuracy could be improved as much as 7.5 W/8)~2 at maximum when the thermal directionality was considered in the SULR estimation.Based on the combined algorithm,the direct physical algorithm was further proposed to extend the applicability from vegetated surfaces to all land surface types.In this algorithm,the relationship between the band-effective radiance and the wide-band radiance was regressed.Combined with the kernel-driven model,the SULR was estimated directly from the surface-leaving radiance.In such a method,the estimation process was simplified,thereby avoiding the error accumulation caused by different input variables.The estimation was also validated using the AWS measurements in the HiWATER experiment.The results showed that the direct physical method was applicable to different land surface types and the estimation accuracy in the direct physical method could be improved by about 0.6 W/8)~2compared with the combined method.At the satellite scale,the MYD21 LST and emissivity products were used to investigate the effects of the angular variation of the emissivity on the SULR estimation.The MYD21 LST and emissivity product are produced using the temperature and emissivity separation(TES)algorithm,which estimates the LST and emissivity simultaneously without a priori of the emissivity.Thus,the estimated LST and emissivity were assumed to view zenith angle(VZA)dependent.A look-up table(LUT)containing directional emissivity in different seasons was built over different land surface types.And the directional wide-band emissivity was extrapolated using the regressed quadratic polynomial.Then the hemispheric broadband emissivity was calculated and used in the SULR estimation.The results showed that the angular variations of the band-effective and wide-band emissivity were both very notable,decreasing with the view zenith angle(VZA)increasing.Compared with the emissivity differences of about 0.04 at Band 29 between nadir and large VZAs,the wide-band emissivity differences between nadir and large VZAs were usually limited to 0.01.The coefficients of determination for the regressed quadratic polynomial at different land surface types were mostly above 0.9.Compared with the SULR estimated using the directional emissivity,the accuracy of the ones estimated using the hemispheric emissivity improved.However,the RMSE differences were limited to 1 W/m~2 over vegetated surface.While over barren soil/sparsely vegetated surface,the difference could reach over 2 W/m~2.Compared with the directionality of emissivity,the angular variation of effective temperature played a more important role in the SULR estimation,which was reflected in the fact that the accuracy of estimated SULR in the day was much higher than that at night.