and Weibull distributions, were used to accurately describe NR clutters of some specific scenarios. In 1981, Ward proposed the product model [14], a turning point in the study of SAR statistical modeling. The product model can be regarded as a generalization of the Rayleigh speckle model. Its derivation is much simpler than the Therefore, based on the random walk model [8], the nonRayleigh (NR) speckle model [9] was proposed, leading to more expressive models such as the K distribution [10] and incoherent scatterer sum (ISS) models [11]-[13], which were used in SAR clutter simulation. Also, empirical probability distributions with more parameters, such as the log-normal Physical Information Maxwell's Equations Coherent Scatterer Model Empirical Model Experimental Electromagnetics (Before 1864) Classical Electromagnetics (1864-1950) Computational Electromagnetics (After 1950) Nonparametric Statistical Model Finite-Mixture Statistical Model Single EmpiricalDistribution Model Rayleigh Speckle Model ISS Model GCL Theorem Model Product Model NR Speckle Model Numerical Method High-/Low-FrequencyApproximation Method Exact Analytical Method High-/Low-FrequencyApproximation Method Exact Analytical Method Faraday's Theorem Coulomb's Law Ampere's Theorem NR Speckle Single-Pixel Statistical Model Spatial-Correlation Model EM Method Statistical Method FIGURE 2. Details of the two main categories of SAR image analysis. Physical Meaning 1960s to the Present 1980s to the Present NR Speckle Model Product Model Early 21st Century to the Present 1960s to the Present NR Speckle GCL Theorem Model ISS Model 1950s to the Present Rayleigh Speckle Model Application Scope 1960s to the Present Finite-Mixture Statistical Model 1990s to the Present Nonparametric Statistical Model 1990s to the Present Mo de l Co mp lex ity Single EmpiricalDistribution Model FIGURE 3. The development of SAR data statistical models. 84 IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE MARCH 2021