Upward looking sonar (ULS) systems are commonly used in ice-covered environments to evaluate sea-ice draft and related characteristics (such as the ice thickness distribution). The involved backscatter data inversion algorithms are mostly based on estimating the two-way propagation time and using some simplifying assumptions which may result in significant limitations, uncertainties, and errors. This work is aimed to develop an improved remote sensing technique of sea-ice characterization based on a more detailed, physics-based analysis of the backscatter intensity time series, the echo shape. The underlying model accounts for two major mechanisms of high-frequency acoustic backscatter, the ice-water interface roughness and volume heterogeneity of the ice layer, provides an estimate of their relative contributions, and includes the echo parameterization in terms of ULS system characteristics (such as the frequency, transmitted pulse’ duration and shape, and the directivity pattern). Using this model, computer simulations were performed for the ULS received signal time series to provide a sensitivity analysis for various physical properties of the ice. Optimal parameters of the system to enhance the ice-properties’ inversion are suggested which, in particular, may result in a possibility to remotely discriminate between first-year and multi-year ice types. [Work supported by ONR.]