Research On Ice Thickness Inversion Method Based On Polarimetric SAR

Ice flood disaster monitoring has always been an important aspect of flood control.Traditional ice observation methods still have shortcomings in terms of timeliness,coverage,and accuracy.Since the beginning of this century,polarimetric synthetic aperture radar(polarimetric SAR),with its characteristics of all-weather,all-day,and high resolution,has gradually become an important means of ice observation.Currently,a considerable amount of research has achieved phased results in the observation and classification of sea ice.However,the study of ice thickness inversion is limited due to the complexity of ice scattering mechanisms and difficulties in data acquisition,particularly in the case of river ice that contains a large amount of sediment,bubbles,and impurities.In this context,this paper conducts in-depth research on the inversion method of ice thickness based on polarimetric SAR,as follows:1.In response to the problem of low accuracy in thickness inversion using the traditional method of constructing an empirical model based on backscattering coefficients,a hybrid empirical and semi-empirical model for thickness inversion was constructed.For thermodynamic ice thickness inversion,the relationship between the thickness of thermodynamic ice,complex permittivity,and polarization angle was derived,and a semiempirical model based on the polarization angle was constructed.For thickness inversion of water-embedded ice,multiple polarimetric parameters were used for fitting analysis,and two parameters with better fitting results were selected to construct an empirical model for waterembedded ice thickness based on multiple polarimetric parameters.Based on these findings,a classification-based hybrid model for ice thickness inversion was proposed.Firstly,the polarimetric SAR data were used to classify the target study area’s ice conditions,and then the hybrid model was used to invert the thickness of thermodynamic ice and water-embedded ice.The accuracy of the inversion results was verified using measured data,showing improvement compared to traditional methods.2.To address the issue of unsatisfactory accuracy in the empirical model based on polarimetric parameters for water-embedded ice thickness inversion in the aforementioned hybrid model,texture information parameters are introduced.This includes the construction of a linear empirical model based on a single texture information parameter,a linear empirical model based on multiple texture information parameters,and a nonlinear empirical model.Based on this,a texture-based water-embedded ice thickness inversion method is proposed.Firstly,taking advantage of the ability of polarimetric SAR data to preserve image texture well,texture information parameters are extracted using a gray-level co-occurrence matrix.Then,a texture-based empirical model is used to invert the thickness of waterembedded ice.The accuracy of the inversion is validated using measured data,demonstrating that the texture-based empirical model improves the inversion accuracy compared to traditional methods and inversion methods based on polarimetric parameters.