A SAR Image Simulation Method and System for Complex Polar Scenes Based on Sea Ice Parametric Modeling
By using power-law fragmentation and dynamic compression mechanisms to parametrically model sea ice, the problem of inverting sea ice physical parameters under extreme environments was solved, and high-similarity simulation of SAR images of complex polar scenes was achieved, thus improving the cognitive ability of polar SAR images.
Patent Information
- Application Number
- CN202611412936.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-11
- Publication Date
- 2026-11-10
AI Technical Summary
Existing technologies struggle to accurately retrieve sea ice physical parameters in extremely harsh polar environments, and SAR image simulations fail to effectively model the complex topological features and electromagnetic scattering behavior of polar sea ice.
A geometric model based on power-law fragmentation and dynamic compression mechanisms is used to construct a three-dimensional sea ice landscape scene, generate discrete floating ice fields, derive the dielectric anisotropy tensor by combining effective medium theory, and generate fully polarimetric SAR images through temporal coherent accumulation to simulate SAR images of complex polar scenes.
It achieves high similarity simulation of polar SAR images, improves the ability to understand polar SAR images, and can accurately simulate the multi-layered rough topology and electromagnetic scattering characteristics of sea ice.
Smart Images

Figure CN123069788A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of polar sea ice observation technology, and in particular to a method and system for SAR image simulation of complex polar scenes based on sea ice parameterization modeling. Background Technology
[0002] Sea ice plays a crucial role in polar air-sea exchange, climate feedback, maritime operations, and Arctic shipping. As Arctic sea ice evolves into thinner, younger, and more fluid states, the need for monitoring its underlying mechanisms is increasingly urgent. Synthetic Aperture Radar (SAR), due to its independence from sunlight and cloud conditions, its ability to provide wide-area, high-resolution coverage, and its sensitivity to geometric roughness and dielectric differences, has become a core sensing tool for sea ice observation. In operational and research practice, SAR is widely used for ice map support, deformation monitoring, lead detection, and ice edge extraction. Although currently orbiting SAR satellites provide a wealth of observational data, the inversion of sea ice physical parameters (such as ice thickness, ice shape, and concentration) remains a highly ill-posed problem in the extremely harsh polar environment due to the lack of spatiotemporally matched measured sea ice data. In sea ice with significant deformation and spatial inhomogeneity, polarization asymmetry and phase behavior are further influenced by the coupling effects of thermodynamic processes and structural changes. Against this backdrop, SAR image simulation technology has become an indispensable auxiliary tool. SAR image simulation constructs a virtual sea ice environment, simulates the working principle of SAR imaging, and obtains simulated SAR images in this virtual environment, enabling the mapping from sea ice environment to SAR images. Forward simulation not only provides a large-scale training dataset with absolute truth values for deep learning inversion algorithms, but also deepens the physical understanding of the interaction mechanism between specific ice features and microwave signals from a theoretical perspective.
[0003] In a complete SAR simulation chain, scene geometric modeling and electromagnetic scattering calculation are two core components that determine simulation fidelity. Regarding geometric modeling, real polar sea ice is a highly dynamic and spatially heterogeneous medium. Mechanical dynamics (such as compression and shearing) generate complex topological features, including discrete ice floe edges and violently deformed ice ridges; simultaneously, melt pools typically evolve on the sea ice surface in summer, all of which are key landforms dominating SAR backscattering characteristics. However, early studies often represented sea ice as a flat surface or large-scale sinusoidal undulations, failing to reflect the fragmentation characteristics of ice floes and the ice ridges formed by ice floe collisions and compression. In recent years, while some studies have introduced random roughness to simulate ice ridge topography, most have failed to establish a rigorous physical connection between roughness distribution and ice floe particle size distribution and mechanical dynamic processes, making it difficult to reproduce the true ice landscape.
[0004] Significant progress has been made in the physical modeling of sea ice for electromagnetic scattering calculations. Numerous studies have elucidated how composition, salinity, temperature, and layered structure affect the equivalent dielectric properties and microwave scattering behavior of sea ice. Related dielectric and mixing formulas have also provided important physical foundations for the parameterization of effective media. For example, the layered reflectivity of SMRT and the microwave forward propagation model of snow and ice provide important references for mechanism-oriented analysis. This type of work is of great value for mechanistic explanation and parameter sensitivity analysis. However, many models remain at the level of materials, layered structure, or scattering mechanisms, and few studies have been able to deeply couple spatially varying dielectric anisotropy with the macroscopic physical parameters of sea ice. Summary of the Invention
[0005] To address the problem of high similarity simulation of SAR images in complex polar regions, this invention provides a method and system for SAR image simulation of complex polar regions based on sea ice parameterization modeling. Based on a geometric model of power-law fragmentation and dynamic compression mechanisms, it faithfully reconstructs a multi-layered rough topological structure including open water, flat ice, ice ridges, and melt pools. Macroscopic topological features are used to drive the generation of spatially heterogeneous temperature, salinity, and density fields. Furthermore, the dielectric anisotropy tensor constrained by the physical field is derived using effective medium theory. The original echo is generated through temporal coherent accumulation and focused to obtain a fully polarized SAR image. This method enables SAR image simulation of complex polar regions, improving the ability to understand polar SAR images.
[0006] According to the design scheme provided by this invention, on the one hand, a method for SAR image simulation of complex polar scenes based on sea ice parameterization modeling is provided, comprising:
[0007] A three-dimensional landform scene of sea ice covering the simulated area is constructed. Discrete floating ice fields are generated based on power-law distribution and dynamic collision and compression mechanism. The geometric parameters of the stacked terrain and the coupling parameters of the multi-layer interface are obtained based on the discrete floating ice fields. The dynamic collision and compression mechanism uses the superposition density operator to describe the mechanical interaction between floating ice. The discrete floating ice fields are composed of irregular polygonal floating ice entities as basic units to quantitatively characterize the multi-dimensional spatial distribution attributes of sea ice in the target sea area. The multi-layer interface includes the air-snow interface, the snow-ice interface, and the ice-water interface.
[0008] The spatial heterogeneous physical properties of sea ice are parameterized, and a local parameter mapping relationship driven by macroscopic geometry is established. Polarization scattering data corresponding to each resolution cell is obtained through multi-interface polarization scattering calculation. The local parameters include temperature parameters, salinity parameters, and density parameters. The mapping relationship maps the local parameters to a linear combination of a reference value and a multidimensional spatial perturbation field. The polarization scattering data includes the interface dielectric anisotropy tensor and polarization components characterizing the birefringence effect inside sea ice.
[0009] Fully polarimetric SAR coherent echo data is generated based on polarimetric scattering data, and fully polarimetric sea ice SAR simulation images are obtained through imaging focusing processing.
[0010] As a method for SAR image simulation of complex polar scenes based on sea ice parameterization modeling in this invention, a three-dimensional landform scene of sea ice covering the area to be simulated is further constructed, and a discrete ice field is generated using a stacking density operator, comprising:
[0011] Set the simulation scene size to match the target observation scale, generate a regular two-dimensional grid covering the area to be simulated, and keep the grid spatial resolution consistent with the ground resolution of the SAR simulation. Each grid cell corresponds to an independent geomorphic attribute cell.
[0012] Discrete ice floes are generated under two-dimensional grid constraints based on power-law distribution and dynamic collision-squeeze mechanism.
[0013] As a method for SAR image simulation of complex polar scenes based on sea ice parameterization modeling in this invention, a discrete ice floe field is further generated based on power-law distribution and dynamic collision-squeeze mechanism, comprising:
[0014] Based on the scenario, the minimum and maximum equivalent radii of the floating ice are determined, and the probability density function of the equivalent radius of the floating ice is constructed.
[0015] Integrating the probability density function yields the cumulative distribution function, and solving for the inverse function of the cumulative distribution function provides the solution.
[0016] A pseudo-random number generator is used to generate independent random numbers, each corresponding to an individual ice floe to be generated. Each random number is substituted into an inverse function to obtain the equivalent radius of the ice floe, and an individual ice floe with a natural shape is generated based on the equivalent radius of the ice floe.
[0017] Using the center of the circumcircle of an individual ice floe as a reference point, the initial circular outline of the ice floe is subjected to controlled random perturbation through a radial harmonic function to generate an irregular edge shape that conforms to the statistical law of sea ice fracture.
[0018] A two-dimensional density distribution function is constructed based on the irregular polygonal contour of a broken ice floe. The normal compressive force and tangential frictional force are determined by the density integral of the overlapping area of two ice floes to simulate the collision, compression and friction interaction process between ice floes.
[0019] As a method for SAR image simulation of complex polar scenes based on sea ice parameterization modeling in this invention, further, it obtains the geometric parameters of ice ridge stacking terrain and multi-layer interface coupling parameters based on discrete ice floes, including:
[0020] Based on the value of the superposition density operator, the simulation scene is divided into open water, flat ice area and ice ridge area;
[0021] The micro-topography of the ice ridge region is modeled using the isotropic exponential power spectrum, and the correlation and statistical self-similarity of the ice ridge topography in space are ensured by the correlation length. The random undulation field is synthesized in the spatial domain by using the two-dimensional inverse fast Fourier transform, so as to determine the random morphological perturbation based on the random undulation field, and the height field of the ice ridge region is obtained by combining the dynamic increment.
[0022] Based on the pre-constructed three-dimensional landform parameter field of sea ice, the height fields of the air-snow interface, snow-ice interface and ice-water interface are independently constructed on a regular two-dimensional grid that matches the ground resolution of SAR simulation, so as to simulate the roughness differences of different layers in real sea ice.
[0023] As a method for SAR image simulation of complex polar scenes based on sea ice parameterization modeling in this invention, the simulation scene is further divided into open water, flat ice area, and ice ridge area according to the value of the stacking density operator, including:
[0024] If the superposition density operator is zero, the corresponding scene will be classified as open water.
[0025] If the stacking density operator is set to 1, the corresponding scene will be classified as a flat ice zone;
[0026] If the superposition density operator has a value greater than 1, the corresponding scene will be classified as an ice ridge region.
[0027] As a method for SAR image simulation of complex polar scenes based on sea ice parameterization modeling in this invention, the height fields of the air-snow interface, the snow-ice interface, and the ice-water interface are further constructed independently, including:
[0028] A two-dimensional distribution field of snow layer thickness is superimposed on a two-dimensional field of height on the upper surface of the ice layer, and a Gaussian random field driven by the snow surface roughness parameter is superimposed to generate a three-dimensional height field of the air-snow interface.
[0029] A Gaussian random field driven by the roughness parameters inside the ice layer is superimposed on the two-dimensional height field of the upper surface of the ice layer to generate a three-dimensional height field of the snow-ice interface.
[0030] Subtract the two-dimensional distribution field of ice thickness from the two-dimensional field of the height of the upper surface of the ice layer, and then superimpose the Gaussian random field driven by the roughness parameter of the bottom of the ice layer to generate the three-dimensional height field of the ice-water interface.
[0031] Among them, the three base planes satisfy the layered distribution constraints of air, snow, ice and water from top to bottom, and the grid of the height field of all interfaces is consistent with the ground resolution grid of SAR simulation.
[0032] As a method for SAR image simulation of complex polar scenes based on sea ice parameterization modeling in this invention, before obtaining the polarization scattering data corresponding to each resolution cell through multi-interface polarization scattering solution, it further includes:
[0033] Using the sea ice dielectric model, with the temperature and salinity profile within each resolution cell as input, the multiphase volume composition of the snow layer, ice layer and water phase is calculated layer by layer. The isotropic reference node constant is solved, and the dielectric anisotropy factor and second-order dielectric tensor are calculated based on the statistical characteristics of the directional arrangement of water inclusions inside the seawater. The calculation granularity of all dielectric parameters is completely aligned with the ground resolution grid of the SAR simulation.
[0034] Furthermore, this invention also provides a SAR image simulation system for complex polar scenes based on sea ice parametric modeling, comprising: a sea ice topography modeling module, a physical heterogeneity modeling module, and a SAR simulation module, wherein...
[0035] The sea ice landform modeling module is used to construct a three-dimensional landform scene of sea ice covering the area to be simulated. It generates discrete ice floes based on power-law distribution and dynamic collision and compression mechanism, and obtains the geometric parameters of the ice ridge stack and the coupling parameters of the multi-layer interface based on the discrete ice floes. The dynamic collision and compression mechanism uses the superposition density operator to describe the mechanical interaction between ice floes. The discrete ice floes are composed of irregular polygonal ice floes as basic units to quantitatively characterize the multi-dimensional spatial distribution attributes of sea ice in the target sea area. The multi-layer interface includes the air-snow interface, the snow-ice interface, and the ice-water interface.
[0036] The physical heterogeneity modeling module is used to parametrically characterize the spatial heterogeneous physical properties of sea ice, establish a local parameter mapping relationship driven by macroscopic geometry, and obtain polarization scattering data corresponding to each resolution cell through multi-interface polarization scattering solution. The local parameters include temperature parameters, salinity parameters and density parameters. The mapping relationship is to map the local parameters to a linear combination of the reference value and the multidimensional spatial perturbation field. The polarization scattering data includes: the interface dielectric anisotropy tensor and polarization components characterizing the birefringence effect inside sea ice.
[0037] The SAR simulation module is used to generate fully polarimetric SAR coherent echo data based on polarimetric scattering data, and to obtain fully polarimetric sea ice SAR simulation images through imaging focusing processing.
[0038] The beneficial effects of this invention are:
[0039] This invention constructs a three-layer geometric model based on power-law fragmentation and collision compression mechanisms, and further evolves ice ridge stacking terrain and multi-layer rough interfaces including air-snow-ice-water. It couples spatially heterogeneous temperature, salinity, density, and dielectric anisotropy physical fields. Under strict geometric and physical constraints, the surface element method is used to calculate the coherent polarization scattering response of multi-layer surfaces and bulk media in detail. It can achieve fully polarized coherent modeling of air-snow surfaces, snow-ice interfaces, ice-water interfaces, and ice volume scattering. Using time-domain pulse-by-pulse technology to generate radar echoes, it can realize SAR image simulation of complex polar scenes and improve the ability to understand polar SAR images. Attached Figure Description
[0040] Figure 1 This is a schematic diagram of the SAR image simulation process for a complex polar scene based on sea ice parameterization modeling in the embodiment.
[0041] Figure 2 This is a schematic diagram of the principle framework of the sea ice parametric modeling and fully polarimetric SAR imaging simulation algorithm in the embodiment;
[0042] Figure 3 This is a schematic diagram of the RD imaging algorithm in the embodiment;
[0043] Figure 4 This is a schematic diagram illustrating the qualitative comparison results between simulated and observed sea ice geometry in the embodiments;
[0044] Figure 5 This is a schematic diagram of the simulated fully polarimetric SAR intensity map in the embodiment. Detailed Implementation
[0045] To make the objectives, technical solutions, and advantages of this invention clearer and more understandable, the invention will be further described in detail below with reference to the accompanying drawings and technical solutions.
[0046] To achieve high similarity simulation of SAR images for complex polar scenes, the embodiments of this invention are described below. Figure 1 As shown, a method for SAR image simulation of complex polar scenes based on sea ice parameterization modeling is provided, including:
[0047] S101. Construct a three-dimensional landform scene of sea ice covering the area to be simulated. Generate a discrete floating ice field based on power-law distribution and dynamic collision compression mechanism. Obtain the geometric parameters of the ice ridge stack and the coupling parameters of the multi-layer interface based on the discrete floating ice field. The dynamic collision compression mechanism uses the superposition density operator to describe the mechanical interaction between floating ice. The discrete floating ice field is composed of irregular polygonal floating ice entities as basic building blocks to quantitatively characterize the multi-dimensional spatial distribution attributes of sea ice in the target sea area. The multi-layer interface includes the air-snow interface, the snow-ice interface, and the ice-water interface.
[0048] Specifically, a simulation scene size matching the target observation scale is set, and a regular two-dimensional grid covering the area to be simulated is generated. The grid spatial resolution is consistent with the ground resolution of the SAR simulation, and each grid cell corresponds to an independent geomorphic attribute cell. Based on the power-law distribution and dynamic collision and compression mechanism, a discrete floating ice field is generated under the constraint of the two-dimensional grid.
[0049] Among them, the discrete ice floe field generated based on the power-law distribution and dynamic collision compression mechanism can be designed to include:
[0050] Based on the scenario, the minimum and maximum equivalent radii of the floating ice are determined, and the probability density function of the equivalent radius of the floating ice is constructed.
[0051] Integrating the probability density function yields the cumulative distribution function, and solving for the inverse function of the cumulative distribution function provides the solution.
[0052] A pseudo-random number generator is used to generate independent random numbers, each corresponding to an individual ice floe to be generated. Each random number is substituted into an inverse function to obtain the equivalent radius of the ice floe, and an individual ice floe with a natural shape is generated based on the equivalent radius of the ice floe.
[0053] Using the center of the circumcircle of an individual ice floe as a reference point, the initial circular outline of the ice floe is subjected to controlled random perturbation through a radial harmonic function to generate an irregular edge shape that conforms to the statistical law of sea ice fracture.
[0054] A two-dimensional density distribution function is constructed based on the irregular polygonal contour of a broken ice floe. The normal compressive force and tangential frictional force are determined by the density integral of the overlapping area of two ice floes to simulate the collision, compression and friction interaction process between ice floes.
[0055] Based on the values of the stacking density operator, the simulation scene can be divided into open water, flat ice area, and ice ridge area. The micro-topography of the ice ridge area is modeled using the isotropic exponential power spectrum, and the correlation length is used to ensure the spatial correlation and statistical self-similarity of the ice ridge topography. The random undulation field is synthesized in the spatial domain using the two-dimensional inverse fast Fourier transform, and the random morphological perturbation is determined based on the random undulation field. The height field of the ice ridge area is obtained by combining the dynamic increment. Based on the pre-constructed three-dimensional landform parameter field of sea ice, the height fields of the air-snow interface, snow-ice interface, and ice-water interface are independently constructed on a regular two-dimensional grid that matches the ground resolution of the SAR simulation, so as to simulate the roughness differences of different layers in real sea ice.
[0056] The fragmentation process of sea ice under dynamic action exhibits significant self-similarity characteristics, and its grain size distribution follows a power law.
[0057] Let the scene size be The probability density function of the equivalent radius r of the floating ice is:
[0058]
[0059] Among them, the power exponent is taken from typical observations. Generated by sampling using an inverse function. Individual ice floes are arranged in descending order to prioritize the placement of larger ice floes, simulating their spatial dominance. To characterize the random fracturing characteristics of natural sea ice edges, a radial harmonic perturbation model is introduced:
[0060]
[0061] Among them, amplitude component follow The attenuation law ensures the continuity of the contour; low-order harmonics define the macroscopic morphology; and high-order perturbations contribute to the edge microstructure, conforming to the statistical laws of the edge morphology of naturally broken ice. Phase Randomization ensures the unique shape of each ice floe.
[0062] To accurately simulate the overlapping and ridge formation process of sea ice under dynamic compression, this invention introduces a superposition density operator. (Overlap Density Operator). use Describe the mechanical interactions between ice floes, based on Based on the value of , the scene is divided into three typical terrain units:
[0063] (a) Open Lead: meets the requirements , indicating a dynamic gap without ice cover;
[0064] (b) Level Ice: meets the following requirements This corresponds to one-year ice (FYI) that has not undergone significant deformation.
[0065] (c) Pressure Ridges: meet the following requirements It simulates the bulging and stacking of two or more ice floes under the pressure of collision, and the higher the C value, the greater the dynamic pressure intensity.
[0066] Ice ridge surfaces exhibit highly non-stationary random undulations. To simulate this multi-scale roughness, this invention employs isotropic exponential power spectra to stochastically model the micro-topography of the ridge region. Its power spectral density function... Defined as:
[0067]
[0068] This spectral model uses relevant length This ensures the spatial correlation and statistical self-similarity of the ice ridge topography. A normalized random undulation field with specific statistical characteristics is synthesized in the spatial domain using a two-dimensional inverse fast Fourier transform (IFFT). :
[0069]
[0070] in, This represents the normalized standard deviation.
[0071] Ultimately, the height field of the glacial ridge region It is formed by the linear coupling of deterministic dynamic increments and stochastic morphological perturbations, and its physical model is expressed as:
[0072]
[0073] In the formula, This represents the average height gain resulting from a single overlap. This is the root mean square height (RMSheight) of the ridge roughness. This formula is derived from... This achieves physical constraints on the micro-roughness strength under the macro-dynamic background.
[0074] During the thermodynamic ablation stage, large-scale ice floes often develop melt pools. This invention establishes a set of melt pool formation criteria based on scale dependence and spatial exclusivity. Let... The Boolean criterion for generating the meltwater pool is the radius of the floating ice. for:
[0075]
[0076] Melt pools only form inside floating ice floes with a radius exceeding a threshold and located in flat ice zones, in accordance with the physical observation facts of "ice ridge drainage and low-lying areas accumulating water." The geometric model of the melt pool is a circular depression with a fixed depth of 0.1m relative to the sea ice freeboard height, and its dielectric constant is calculated using electromagnetic scattering based on a freshwater model.
[0077] The sea ice scene comprises three key interfaces: the air-snow interface (the upper surface of the snow layer), the snow-ice interface (the upper surface of the ice layer), and the ice-water interface (the lower surface of the ice layer). The height fields of each interface are constructed independently to simulate the roughness differences between different layers in real sea ice.
[0078] Air-snow interface height field From the height of the upper surface of the ice layer Snow layer thickness and snow surface roughness composition:
[0079]
[0080] snow surface roughness An exponential spectrum was used to simulate the subtle fluctuations caused by windblown snow.
[0081] Snow-ice interface height field Independent of the snow layer, it consists only of the height of the upper surface of the ice layer and its internal roughness:
[0082]
[0083] in This reflects the small-scale irregularities caused by the internal crystal structure of the ice layer. The independence of this interface decouples the scattering of radar waves at the snow-ice interface from that at the air-snow interface, which is more consistent with the characteristics of actual layered media.
[0084] Ice-water interface height field Based on ice thickness With bottom roughness generate:
[0085]
[0086] All height fields are smoothed using Gaussian filtering to avoid jagged boundaries caused by surface pixel discretization, while preserving large-scale terrain features.
[0087] S102. Parametrically characterize the spatial heterogeneous physical properties of sea ice, establish a local parameter mapping relationship driven by macroscopic geometry, and obtain polarization scattering data corresponding to each resolution cell through multi-interface polarization scattering calculation. The local parameters include temperature parameters, salinity parameters and density parameters. The mapping relationship is to map the local parameters to a linear combination of the reference value and the multidimensional spatial perturbation field. The polarization scattering data includes: the interface dielectric anisotropy tensor and polarization components characterizing the birefringence effect inside sea ice.
[0088] In traditional sea ice SAR simulation studies, ice layers are usually simplified as a medium with uniform physical parameters, which cannot explain the complex textures and polarization anomalies caused by undulations in high-resolution SAR images. In this embodiment, a set of geometric feature mapping physical parameters is established, and the physical parameters drive the forward cascade model of the polarization dielectric tensor, realizing a high-fidelity spatial heterogeneous distribution of temperature, salinity, density, and dielectric anisotropy within sea ice.
[0089] The thermodynamic and kinetic properties of natural sea ice are highly dependent on its macroscopic geometry. In this embodiment, a multivariate spatial modulation model is used to measure the snow layer of sea ice (…). Ice layer ( physical parameter set (Representing temperature, salinity, and density respectively) are parameterized as a linear combination of the baseline values and the multidimensional spatial perturbation field:
[0090]
[0091] In the formula, This is a physical baseline value set based on typical Arctic sea ice observation data; A stationary two-dimensional exponential random field to characterize natural thermal fluctuations; and These are the normalized ice floe edge intensity mask and ice ridge stacking intensity mask extracted from the geometric model, respectively. These are the physical modulation weights corresponding to each effect. The specific settings and physical mechanisms of each parameter are as follows:
[0092] a) Spatial heterogeneous temperature field
[0093] The reference temperatures for snow and ice layers are set as typical winter stratification profile values. and .
[0094]
[0095] Random Field (Relevant length) The positive weights reflect large-scale thermal fluctuations caused by cloud cover or snow thickness. The weights for the edge region and ice ridge region are set to negative values. The edge of the floating ice is usually thinner, and the large heat conduction gradient leads to lower temperatures; while the ice ridge region is more exposed to the extremely cold atmosphere due to orographic uplift, resulting in localized strong cooling.
[0096] b) Spatial heterogeneous salinity field :
[0097] The baseline values for the salinity field (unit: ppt) were taken as follows: and .
[0098]
[0099] Ice ridges are given extremely high positive weights ( This is because during the dynamic compression and stacking process, the brine channels inside the ice layer deform and become blocked, resulting in a strong "brine trapping effect." The small positive weight in the edge region simulates the localized increase in salinity caused by brine seepage from the fault zone. In addition, the ice layer salinity in the freshwater melt pool area is forcibly reduced to 2.8 ppt due to meltwater dilution.
[0100] c) Spatial heterogeneous density field :
[0101] Density field (unit: kg / m3) reference value and .
[0102]
[0103] The weighting signs at ice ridges exhibit inverse characteristics between snow and ice layers. The weighting of snow ridges is positive (+35), attributed to mechanical compaction caused by wind-blown snow; while the weighting of ice ridges is negative (-22), because the disordered stacking of ice blocks introduces a large number of macroscopic pores, significantly reducing the equivalent bulk density.
[0104] Before multi-interface polarization scattering calculation, the sea ice dielectric model can be used to calculate the multiphase volume composition of snow, ice and water layers layer by layer, with the temperature and salinity profile in each resolution cell as input. The isotropic reference node constants can be solved, and the dielectric anisotropy factor and second-order dielectric tensor can be calculated based on the statistical characteristics of the directional arrangement of water inclusions in the seawater. The calculation granularity of all dielectric parameters is completely aligned with the ground resolution grid of the SAR simulation.
[0105] Based on heterogeneous physical parameter fields, an equivalent polarization complex permittivity tensor is constructed for the microwave band. Sea ice dielectric models are typically considered as multiphase mixed media (pure ice, air pores, and brine inclusions).
[0106] a) Multiphase volumetric composition calculation
[0107] The snow layer is considered a three-phase mixture of air, pure ice, and trace amounts of brine. Its pure ice volume fraction is... The ice layer consists of a pure ice matrix, brine, and pores, with its porosity calculated as follows: Volume fraction of brine within the ice layer Strictly dependent on local temperature and salinity:
[0108]
[0109] In the formula These are the empirical polynomial coefficients that vary with temperature.
[0110] b) Isotropic reference permittivity
[0111] The isotropic equivalent dielectric constant is calculated using the Looyenga mixing formula, which has high accuracy for random multiphase media in the microwave band. :
[0112]
[0113] For ice layers, the basic components include:
[0114] pure ice : Calculated as ;
[0115] brine The Debye relaxation model is adopted, and the loss of ionic conductivity caused by high salinity is considered:
[0116]
[0117] In the formula All and The function, For radar operating frequency, It represents the vacuum permittivity.
[0118] c) Dielectric anisotropy factor and tensor decomposition
[0119] Saltwater inclusions within natural sea ice exhibit a significant vertical orientation along the thermodynamic gradient, leading to anisotropy in polarization scattering. This invention introduces a dielectric anisotropy factor. And parameterize it as a function of volumetric components and normalized environmental parameters:
[0120]
[0121] In the formula This is a normalized parameter. Its physical meaning is: the volume fraction of salt water. The contribution to anisotropy is dominant. Subsequently, the relationship between the anisotropy factor and the polarization complex permittivity is:
[0122]
[0123] In the formula This is the isotropic equivalent dielectric constant. This relationship ensures... The polarization dielectric constant of VV is higher than that of HH, which is consistent with sea ice observations.
[0124] For the three interfaces, the 2×2 polarization scattering matrix of each interface is obtained. The primary and secondary reflection scattering components of all interfaces are correlated and superimposed to obtain the total surface scattering matrix. For each medium layer, based on the dense medium radiation propagation theory, the volume scattering coefficient, extinction coefficient, and phase matrix of each medium layer are calculated. By solving the vector radiative transfer equation, the volume scattering Mueller matrix of the snow and ice layers is obtained, and the coupled components of surface scattering and volume scattering are calculated. The surface scattering matrix, volume scattering matrix, and coupled scattering components are uniformly transformed to the same polarization basis, and the co-polarized and cross-polarized components are correlated and superimposed to obtain the total 2×2 polarization scattering matrix of each resolution cell.
[0125] S103. Generate fully polarimetric SAR coherent echo data based on polarimetric scattering data, and obtain fully polarimetric sea ice SAR simulation images through imaging focusing processing.
[0126] SAR synthesizes a virtual large-aperture antenna by utilizing the relative motion between the radar platform and the sea ice scene. It achieves high range resolution using pulse compression and high azimuth resolution using Doppler spectral analysis, ultimately mapping the two-dimensional polarized scattering field into a fully polarized complex image. The radar operates at a speed... Operation, transmitting linear frequency modulated (LFM) signals. For the four polarization channels HH, HV, VH, and VV, the first... The coherent echo of each pulse is:
[0127]
[0128] in The simulation incorporates a two-way time delay. A geometric shadowing criterion is introduced: if the dot product of the surface element's normal vector and the radar line-of-sight vector is greater than zero, the surface element is considered invisible. Multi-threaded parallel processing is employed to address the computational challenges posed by millions of surface elements.
[0129] The principle framework of the sea ice parametric modeling and fully polarimetric SAR imaging simulation algorithm in this case is as follows: Figure 2 As shown, firstly, a discrete ice field is generated based on a power-law distribution and dynamic collision compression mechanism, and further evolves into ice ridge stacking topography and a multi-layered rough interface containing air-snow-ice-water. Based on the sea ice geometric model, a mapping relationship is established between parameters such as local temperature, salinity, and density driven by macroscopic geometry, and the dielectric anisotropy tensor and its horizontal and vertical polarization components characterizing the birefringence effect inside the sea ice are analytically solved. Subsequently, under strict geometric and physical constraints, the coherent polarization scattering response of the multi-layered surface and bulk medium is precisely calculated using the element-based method, and radar echoes are generated using time-domain pulse-by-pulse technology. Finally, simulated synthetic aperture radar (SAR) images are obtained through imaging processing.
[0130] The RD imaging algorithm is used to process the time-domain echo. The imaging process of SAR echo signals can be equivalent to the process of compressing and focusing the echo signals along the azimuth and range directions, such as... Figure 3 As shown, SAR echo imaging is a two-dimensional deconvolution process for reconstructing the electromagnetic scattering coefficients of a target; that is, a SAR image is a mapping of the target's electromagnetic scattering coefficients.
[0131] Furthermore, based on the above method, this embodiment of the invention also provides a SAR image simulation system for complex polar scene based on sea ice parametric modeling, comprising: a sea ice topography modeling module, a physical heterogeneity modeling module, and a SAR simulation module, wherein,
[0132] The sea ice landform modeling module is used to construct a three-dimensional landform scene of sea ice covering the area to be simulated. It generates discrete ice floes based on power-law distribution and dynamic collision and compression mechanism, and obtains the geometric parameters of the ice ridge stack and the coupling parameters of the multi-layer interface based on the discrete ice floes. The dynamic collision and compression mechanism uses the superposition density operator to describe the mechanical interaction between ice floes. The discrete ice floes are composed of irregular polygonal ice floes as basic units to quantitatively characterize the multi-dimensional spatial distribution attributes of sea ice in the target sea area. The multi-layer interface includes the air-snow interface, the snow-ice interface, and the ice-water interface.
[0133] The physical heterogeneity modeling module is used to parametrically characterize the spatial heterogeneous physical properties of sea ice, establish a local parameter mapping relationship driven by macroscopic geometry, and obtain polarization scattering data corresponding to each resolution cell through multi-interface polarization scattering solution. The local parameters include temperature parameters, salinity parameters and density parameters. The mapping relationship is to map the local parameters to a linear combination of the reference value and the multidimensional spatial perturbation field. The polarization scattering data includes: the interface dielectric anisotropy tensor and polarization components characterizing the birefringence effect inside sea ice.
[0134] The SAR simulation module is used to generate fully polarimetric SAR coherent echo data based on polarimetric scattering data, and to obtain fully polarimetric sea ice SAR simulation images through imaging focusing processing.
[0135] To verify the effectiveness of this solution, the following explanation is based on experimental data:
[0136] Computer configuration: Operating system: Windows 11 64-bit, Processor: Intel(R) Core(TM) i7-8750H 2.20GHz, Memory: 32 GB, Graphics card: NVIDIA Quadro P1000, Algorithm implementation platform: MATLAB R2021b software.
[0137] Experimental data description: The simulated sea ice scene size was set to 120m × 120m, with a spatial sampling interval of 1m. The geometric parameters and SAR imaging parameters of the sea ice scene are shown in Table 1. The proportions of ice ridges, melt pools, open water, and flat ice are naturally determined by power-law floating ice generation, the overlap density operator, and the melt pool criterion.
[0138]
[0139] Experimental results are as follows Figure 4 and 5 As shown. Figure 4 The simulated sea ice geometry was compared with the ice morphology observed by UAVs. (a) Simulated sea ice geometry generated by the proposed framework, (b) Sea ice geometry observed by UAVs. Figure 5 This is a simulated intensity map of a fully polarimetric synthetic aperture radar (SAR), where (a)-(d) represent the four polarizations: HH, VV, VH, and HV, respectively.
[0140] The experimental data above demonstrate that the proposed solution, by constructing a three-layer geometric model based on power-law fragmentation and collision compression mechanisms and coupling it with the temperature, salinity, density, and dielectric anisotropy physical fields of spatial heterogeneity, achieves fully polarimetric coherent modeling and SAR imaging simulation of air-snow surface, snow-ice interface, ice-water interface, and ice volume scattering. This enables SAR image simulation of complex polar scenes, thereby improving the ability to understand polar SAR images.
[0141] Unless otherwise specifically stated, the relative steps, numerical expressions, and values of the components and steps described in these embodiments do not limit the scope of the invention.
[0142] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple; relevant parts can be referred to the method section.
[0143] The units and method steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of each example have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations are not considered to be beyond the scope of this invention.
[0144] Those skilled in the art will understand that all or part of the steps in the above methods can be implemented by a program instructing related hardware, and the program can be stored in a computer-readable storage medium, such as a read-only memory, a disk, or an optical disk. Optionally, all or part of the steps in the above embodiments can also be implemented using one or more integrated circuits. Accordingly, each module / unit in the above embodiments can be implemented in hardware or as a software functional module. This invention is not limited to any particular combination of hardware and software.
[0145] Finally, it should be noted that the above-described embodiments are merely specific implementations of the present invention, used to illustrate the technical solutions of the present invention, and not to limit it. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments within the technical scope disclosed in the present invention, or make equivalent substitutions for some of the technical features; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.
Claims
1. A method for SAR image simulation of complex polar scenes based on sea ice parametric modeling, characterized in that, Include: A three-dimensional landform scene of sea ice covering the simulated area is constructed. Discrete floating ice fields are generated based on power-law distribution and dynamic collision and compression mechanism. The geometric parameters of the stacked terrain and the coupling parameters of the multi-layer interface are obtained based on the discrete floating ice fields. The dynamic collision and compression mechanism uses the superposition density operator to describe the mechanical interaction between floating ice. The discrete floating ice fields are composed of irregular polygonal floating ice entities as basic units to quantitatively characterize the multi-dimensional spatial distribution attributes of sea ice in the target sea area. The multi-layer interface includes the air-snow interface, the snow-ice interface, and the ice-water interface. The spatial heterogeneous physical properties of sea ice are parameterized, and a local parameter mapping relationship driven by macroscopic geometry is established. Polarization scattering data corresponding to each resolution cell is obtained through multi-interface polarization scattering calculation. The local parameters include temperature parameters, salinity parameters, and density parameters. The mapping relationship maps the local parameters to a linear combination of a reference value and a multidimensional spatial perturbation field. The polarization scattering data includes the interface dielectric anisotropy tensor and polarization components characterizing the birefringence effect inside sea ice. Fully polarimetric SAR coherent echo data is generated based on polarimetric scattering data, and fully polarimetric sea ice SAR simulation images are obtained through imaging focusing processing.
2. The SAR image simulation method for complex polar scenes based on sea ice parameterization modeling according to claim 1, characterized in that, Construct a 3D sea ice topographic scene covering the area to be simulated, and generate discrete ice floes using the stacking density operator, including: Set the simulation scene size to match the target observation scale, generate a regular two-dimensional grid covering the area to be simulated, and keep the grid spatial resolution consistent with the ground resolution of the SAR simulation. Each grid cell corresponds to an independent geomorphic attribute cell. Discrete ice floes are generated under two-dimensional grid constraints based on power-law distribution and dynamic collision-squeeze mechanism.
3. The SAR image simulation method for complex polar scenes based on sea ice parameterization modeling according to claim 1 or 2, characterized in that, Discrete ice floes are generated based on power-law distribution and dynamic collision compression mechanisms, including: Based on the scenario, the minimum and maximum equivalent radii of the floating ice are determined, and the probability density function of the equivalent radius of the floating ice is constructed. Integrating the probability density function yields the cumulative distribution function, and solving for the inverse function of the cumulative distribution function provides the solution. A pseudo-random number generator is used to generate independent random numbers, each corresponding to an individual ice floe to be generated. Each random number is substituted into an inverse function to obtain the equivalent radius of the ice floe, and an individual ice floe with a natural shape is generated based on the equivalent radius of the ice floe. Using the center of the circumcircle of an individual ice floe as a reference point, the initial circular outline of the ice floe is subjected to controlled random perturbation through a radial harmonic function to generate an irregular edge shape that conforms to the statistical law of sea ice fracture. A two-dimensional density distribution function is constructed based on the irregular polygonal contour of a broken ice floe. The normal compressive force and tangential frictional force are determined by the density integral of the overlapping area of two ice floes to simulate the collision, compression and friction interaction process between ice floes.
4. The SAR image simulation method for complex polar scenes based on sea ice parameterization modeling according to claim 1, characterized in that, Based on discrete ice floe fields, the geometric parameters of ice ridge stacking terrain and multi-layer interface coupling parameters are obtained, including: Based on the value of the superposition density operator, the simulation scene is divided into open water, flat ice area and ice ridge area; The micro-topography of the ice ridge region is modeled using the isotropic exponential power spectrum, and the correlation and statistical self-similarity of the ice ridge topography in space are ensured by the correlation length. The random undulation field is synthesized in the spatial domain by using the two-dimensional inverse fast Fourier transform, so as to determine the random morphological perturbation based on the random undulation field, and the height field of the ice ridge region is obtained by combining the dynamic increment. Based on the pre-constructed three-dimensional landform parameter field of sea ice, the height fields of the air-snow interface, snow-ice interface and ice-water interface are independently constructed on a regular two-dimensional grid that matches the ground resolution of SAR simulation, so as to simulate the roughness differences of different layers in real sea ice.
5. The SAR image simulation method for complex polar scenes based on sea ice parameterization modeling according to claim 4, characterized in that, Based on the values of the superposition density operator, the simulation scene is divided into open water, flat ice areas, and ice ridge areas, including: If the superposition density operator is zero, the corresponding scene will be classified as open water. If the stacking density operator is set to 1, the corresponding scene will be classified as a flat ice zone; If the superposition density operator has a value greater than 1, the corresponding scene will be classified as an ice ridge region.
6. The SAR image simulation method for complex polar scenes based on sea ice parameterization modeling according to claim 1, characterized in that, The height fields of the air-snow interface, the snow-ice interface, and the ice-water interface are constructed independently, including: A two-dimensional distribution field of snow layer thickness is superimposed on a two-dimensional field of height on the upper surface of the ice layer, and a Gaussian random field driven by the snow surface roughness parameter is superimposed to generate a three-dimensional height field of the air-snow interface. A Gaussian random field driven by the roughness parameters inside the ice layer is superimposed on the two-dimensional height field of the upper surface of the ice layer to generate a three-dimensional height field of the snow-ice interface. Subtract the two-dimensional distribution field of ice thickness from the two-dimensional field of the height of the upper surface of the ice layer, and then superimpose the Gaussian random field driven by the roughness parameter of the bottom of the ice layer to generate the three-dimensional height field of the ice-water interface. Among them, the three base planes satisfy the layered distribution constraints of air, snow, ice and water from top to bottom, and the grid of the height field of all interfaces is consistent with the ground resolution grid of SAR simulation.
7. The SAR image simulation method for complex polar scenes based on sea ice parameterization modeling according to claim 1, characterized in that, Before obtaining the polarization scattering data corresponding to each resolution cell through multi-interface polarization scattering solution, the following is also included: Using the sea ice dielectric model, with the temperature and salinity profile within each resolution cell as input, the multiphase volume composition of the snow layer, ice layer and water phase is calculated layer by layer. The isotropic reference node constant is solved, and the dielectric anisotropy factor and second-order dielectric tensor are calculated based on the statistical characteristics of the directional arrangement of water inclusions inside the seawater. The calculation granularity of all dielectric parameters is completely aligned with the ground resolution grid of the SAR simulation.
8. A SAR image simulation system for complex polar scenes based on sea ice parametric modeling, characterized in that, It includes: a sea ice landform modeling module, a physical heterogeneity modeling module, and a SAR simulation module. The sea ice landform modeling module is used to construct a three-dimensional landform scene of sea ice covering the area to be simulated. It generates discrete ice floes based on power-law distribution and dynamic collision and compression mechanism, and obtains the geometric parameters of the ice ridge stack and the coupling parameters of the multi-layer interface based on the discrete ice floes. The dynamic collision and compression mechanism uses the superposition density operator to describe the mechanical interaction between ice floes. The discrete ice floes are composed of irregular polygonal ice floes as basic units to quantitatively characterize the multi-dimensional spatial distribution attributes of sea ice in the target sea area. The multi-layer interface includes the air-snow interface, the snow-ice interface, and the ice-water interface. The physical heterogeneity modeling module is used to parametrically characterize the spatial heterogeneous physical properties of sea ice, establish a local parameter mapping relationship driven by macroscopic geometry, and obtain polarization scattering data corresponding to each resolution cell through multi-interface polarization scattering solution. The local parameters include temperature parameters, salinity parameters and density parameters. The mapping relationship is to map the local parameters to a linear combination of the reference value and the multidimensional spatial perturbation field. The polarization scattering data includes: the interface dielectric anisotropy tensor and polarization components characterizing the birefringence effect inside sea ice. The SAR simulation module is used to generate fully polarimetric SAR coherent echo data based on polarimetric scattering data, and to obtain fully polarimetric sea ice SAR simulation images through imaging focusing processing.
9. An electronic device, characterized in that, include: At least one processor, and a memory coupled to said at least one processor; The memory stores a computer program that can be executed by the at least one processor to implement the method as described in any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed, enables the implementation of the method as described in any one of claims 1 to 7.
