| Snow cover is an important part of the cryosphere and is one of the most active natural elements of the Earth's surface.Seasonal snow cover is of great significance fo r the ecological security and regio nal coordinated development of arid mountainous areas in the arid and semi-arid regions,as well as the main source of freshwater in the northwest mountainous areas of China.The accumulation and melting of snow,with the storage and release of energy and water,play a very important role in the climate system.Manasi River Basin is located in arid region of Northwest China,which is not only the economic core area of the northern slope of Tianshan Mountains,but also is an ecologically fragile area.Liquid water content(LWC)of snow is the quantity of liquid water filling the pores between snow grains of snow cover.It is an important indicator to characterize snow melting process.The information of its spatial and temporal variations can provide guidance to snowmelt runoff forecast as well as regional climate change research.As the main means of earth observation,remote sensing makes it possible to estimate LWC in large-scale,high frequency,and high accuracy.It is also the only effective means of obtaining the temporal and spatial variation of LWC in harsh mountainous area.Synthetic aperture radar(SAR),with high spatial resolution,which can obtain information of snow layers,plays an important role in the retrieval of LWC.Therefore,it has important significance to the water resource management and regional climate research to retrieve LWC with SAR technology.This study focused on the needs of National Science and Technology Major Project "Snow and ice monitoring and its evaluation based on high-resolution remote sensing data in central Tianshan Mountains in Xinjiang Province"(Grant No.95-Y40B02-9001-13/15-04)and National Natural Science Foundation Project "Joint inversion of snow water equivalent based on SAR and high resolution optical remote sensing"(Grant No.41271353).For retrieving LWC,with the synchronous observation data,the feasibility of LWC retrieval in mountainous area based on C-band Radarsat-2 was discussed,and the problems of model improvement and inversion process optimization in complicated-terrain mountainous area were solved.The main research contents and conclusions are as follows:(1)Microwave properties of the wet snow in the study area.Synchronous observation data was used to choose the most adaptable empirical model to describe the relationship between LWC and dielectric constant for the study area.Based on this,the penetration depth of wet snow was simulated.The results revealed that the penetration capabilities of the microwaves decreased as the LWC increased.At the C-band,the information of 10 cm below the top layer can be acquired when the LWC exceeded 2-3%.Based on the wet snow map extracted from the SAR data,backscattering from the wet snow was decomposed into surface scattering and volume scattering using polarimetric decomposition.It was verified that the snow surface scattering was the dominant component of the wet snow scattering.The results can provide theoretical support for determining influencing factors of the inverse model for calculating LWC of snow surface.(2)Determination of model parameters.The Advanced Integral Equation M ethod(AIEM)model was used to simulate surface scattering of wet snow.The influencing factors on the response of the LWC and backscattering coefficient were determined,which are frequency,polarization mode,local incidence angle,snow suface density,and snow suface roughness.The responses of surface backscattering coefficient at C-band and influencing factors were analyzed.Methods for controlling these factors'influence were proposed.Co-polarization was found to be the optimal polarization mode for inverse modeling of LWC.The ratios of surface scattering in the same phase but polarized in different modes were used,in order to reduce the influence of snow surface roughness.Based on topographical data,the snow surface density was interpolated as an input parameter of the model.Local incident angle,which can be exactly acquired,was used as another input parameter.Research found that the ratios of volume scattering for different polarization modes can be represented as a function of merely snow dielectric constants and local incident angle.Therefore,the input parameters of the inverse model were local incident angle and snow surface density.These can be used as theoretical basis for subsequent model modification.(3)Improvement of inversion model.The range from the synchronous observations and Radarsat-2 were used to build a dataset to simplify the form of surface scattering model.Optimal parameter setting was determined through a comparative analysis,and the value ranges of the coefficients in the simplified surface scattering model were fitted.Combining the simplified surface and volume scattering model,the inverse model for calculating LWC of snow surface in the study area was constructed.An interpolation method combining simulated surface interpolation,residual error interpolation,and topography-based temporal adjustment was presented.The snow surface density parameter was derived.To optimize inversion process and improve the accuracy,the study proposed the idea of dynamic inversion.It refers to a process of the output results can be constrainted by adjusting the snow surface density and model coefficients in each range.These provide technical support to LWC retrieval.(4)Evaluation of inversion results.With the calculated local incidence angle and snow density,the map of LWC in the study area was obtained and the distribution of LWC was discussed.It is found that the LWC were small,mainly in the range of 3?4%,and the average was only 3.57%,on March 19,2014.Elevation and aspect are the main influence topographical factor of its distribution.The average of LWC of lower elevations at south slope was maximum,which was minimum in higher elevations at north slope.The comparison of SAR-retrieved LWC and the ground measurement indicated that the mean absolute error was 0.64%by volume,and the absolute error at 95%confidence interval was 0.88%.The result of LWC with high precision illustrated that the LWC retrieval is feasible and effective based on multi-temporal C-band S AR data and improved model.These results provide practical application to LWC inversion.This study discussed LWC inversion in a mountainous area.The microwave properties of the wet snow in this area were analyzed based on field measurements.The main influencing factors at C-band were determined,and the methods were proposed for suppressing the influence of factors affecting the relationship between LWC and backscattering coefficient.After that,the input parameters of the inverse model were determined,providing theoretical support for subsequent research.In order to improve the applicability of the existing model to the study area,the optimization method of existing model was put forward.The dynamic range of input parameters and model coefficients is obtained.The dynamic control of the inversion process is realized.These achievements have certain theoretical significance,technical innovation,and practical application value. |
PDF Full Text Request