Study On Estimation Methods Of Forest Microwave Transmissivity During Snow Season In Northeast China

Satellite remote sensing observation using microwave radiometry is an effective method for monitoring dynamic changes in the cryosphere,atmosphere,hydrosphere,etc.Continuous long time series of satellite-based passive microwave observations can not only provide reference information on the current state of the environment,but also be used to study the development trends related to climate change.However,one of the main challenges of using satellite-based passive microwave monitoring of snow information is the mixed image problem caused by the low spatial resolution of sensor observations,which usually refers to the spatial heterogeneity within the image due to different land cover types;the second challenge is the effect of forest cover,which significantly attenuates the uplink microwave radiation of snow on the subsurface and adds some thermal radiation to the canopy itself;therefore,a complex microwave radiation model is required.Therefore,a complex microwave radiation model is needed to relate the near-ground measurement parameters(e.g.,brightness temperature,etc.)to the physical properties of the media,such as snow,forest vegetation,and soil layers,in order to provide a reasonable interpretation of the measurement results in nonhomogeneous areas.If the effect of forest vegetation is not taken into account in the algorithmic model,the inverse parameters of snow accumulation in the forest area will produce large errors and uncertainties.Forest canopy microwave transmissivity is a key parameter for describing forest radiation,which not only affects the proportion of microwave radiation penetrating the forest canopy,but also directly relates to the thermal radiation of the forest itself.Numerous studies have shown that it is a crucial parameter for correcting remote sensing observations of land surface parameters in forest areas.Northeast China is not only an important timber production base and food security base in China,but also one of the three major stable snow areas in China,with a forest coverage of about 40%.Paying attention to the information of snow conditions in forest areas is helpful to grasp the climate change dynamics in time,improve the early warning capability of drought and flood disasters as well as provide reference information for the development of climate,hydrology and agroecology.However,the problem of inversion of snow parameters in forest areas has not been well solved.In order to effectively improve the inversion accuracy of snow parameters in forest areas,this thesis starts from the forest vegetation radiative transfer model and focuses on the research objective of how to estimate the parameters of forest transmissivity,with the following research methods,contents and corresponding conclusions:(1)Ground-based microwave radiation observation experiments were carried out in typical forest areas of Northeast China’s Greater and Lesser Xing’an Mountains.The ground-based measurement data set includes radiometric data,environmental element observation data,forest canopy and structural parameters,etc.The forest transmissivity parameters under K-band(18.7 GHz)and Ka-band(36.5 GHz)horizontal and vertical polarization of each forest sample site were obtained through model calculations,and based on the orthorectified The main conclusions include:(1)The satellite bright temperature correlations based on the embedded transmissivity parameters show that there is a body scattering effect in the forest under K-band horizontal polarization,while there is a body scattering effect in the forest under Ka-band dual polarization and K-band vertical polarization.(2)The forest stand volume parameter can be used as an effective parameter to characterize the forest transmissivity in Northeast China,which can provide a basis for further development of transmissivity-storage models for different forest types.Meanwhile,the inversion algorithm of snow depth observed at 5-6 meteorological stations in northeastern forest area was corrected by applying the transmissivity measurement data,which effectively improved the snow depth inversion accuracy,and the RMSE was reduced from 10 cm to 6.7 cm overall,with a reduction of 33%.(2)The relationship between forest transmissivity and stand volume was investigated for three common forest types in northeast China,namely deciduous coniferous forest,deciduous broadleaf forest and evergreen coniferous forest,and an empirical model set of e index was developed for each forest type by optimal fitting,with the applicable frequencies of 18.7 GHz and 36.5 GHz,polarization modes of horizontal and vertical polarization,and an effective range of forest volume from 0 to 300 m3/ha.The newly developed forest transmissivity models show that the asymptotic saturation level of forest transmissivity varies depending on the forest type,and the transmissivity at each frequency and polarization is generally the highest for deciduous broadleaf forests,followed by deciduous coniferous forests,and the lowest for evergreen coniferous forests when the accumulation is large.The lowest transmissivity was observed in the deciduous broadleaf forest.Based on the radiative transfer theory,the results of forward simulation of on-star bright temperature show that the newly developed transmissivity model has good performance and the forward simulation accuracy is between 3-6 K.It is proved that it can be applied as a nested sub-model of the radiative transfer model system for snow accumulation parametric inversion.(3)To further address the limitations of the sample point scale model for image element scale applications,we constructed a satellite image element scale forest transmissivity inversion algorithm based on the long time series SSMIS observed bright temperature data and ERA5-Land temperature element set for five consecutive winters from 2014-2019,based on a simplified radiative transfer model,and applied an iterative optimization method to determine the optimal parameters to produce The forest microwave transmissivity dataset was produced for the northeast region.From the average of the values,the overall K-band transmissivity is higher than Ka-band,and the overall horizontal polarized transmissivity is higher than the vertical polarized transmissivity.The interannual temporal variation is relatively stable,with no significant increase or decrease,and the coefficient of variation of transmissivity in most of the image areas does not exceed 0.15.The spatial distribution shows that the transmissivity at the edge of the forest area is higher than that inside the forest area.From the application point of view,the performance of the pixel-wise transmissivity model developed in this thesis is well tested.Compared with the e-exponential accumulation-transmissivity model(referred to as the Glob Snow model algorithm)used in the Glon Snow 3.0 snow accumulation product,the accuracy is significantly improved in the inversion application of snow depth SD and snow water equivalent SWE.In terms of the index RMSE,the accuracy of SD and SWE inversion is improved by 6.8 cm and 7.75 mm,respectively,with an effective accuracy improvement of 41% and 26%,respectively;in terms of the index Bias,the accuracy of SD and SWE inversion is improved by 84% and 95%,respectively;in addition,the correlation coefficient is also improved.Therefore,the pixel-wise transmissivity model developed in this thesis can provide application value for the operational inversion of snow accumulation parameters by passive microwave satellite remote sensing.In summary,this dissertation research focuses on the ultimate goal of improving the inversion accuracy of forest snow accumulation parameters and the primary goal of forest transmissivity parameter estimation,and carries out a series of basic experimental and theoretical research work.Firstly,based on the ground-based microwave radiation measurement experiment,after a comprehensive investigation and understanding of the microwave radiation transmission characteristics of the coupled forest-snow system,a forest transmissivity-storey function relationship model for different forest types is developed based on the conclusions obtained from the experiment,and further extended from the sample point scale to the image element scale,and an image element-level forest transmissivity model and a series of data sets are developed,which provide a methodological reference for the forest microwave radiation impact correction in forest areas.It also provides data support for the operational inversion algorithm of passive microwave satellite remote sensing snow accumulation parameters in forest areas.In conclusion,this study quantitatively analyzes the factors affecting the radiation transmission process of forest-snow system in terms of mechanism research;in terms of remote sensing application,it develops and verifies the applicability and spatial and temporal generalizability of the pixel-wise transmissivity inversion model.