Impact of Microstructure on an Accurate Snow Scattering Parameterization at Microwave Wavelengths

High frequency microwave instruments are increasingly used to observe ice clouds and snow. These instruments are significantly more sensitive than conventional precipitation radar. This is ideal for analyzing ice-bearing clouds, for ice particles are tenuously distributed and have effective densities that are far less than liquid water. However, at shorter wavelengths, the electromagnetic response of ice particles is no longer solely dependent on particle mass. The shape of the ice particles also plays a significant role. Thus, in order to understand the observations of high frequency microwave radars and radiometers, it is essential to model the scattering properties of snowflakes correctly. Several research groups have proposed detailed models of snow aggregation. These particle models are coupled with computer codes that determine the particles' electromagnetic properties. However, there is a discrepancy between the particle model outputs and the requirements of the electromagnetic models. Snowflakes have countless variations in structure, but we also know that physically similar snowflakes scatter light in much the same manner. Structurally exact electromagnetic models, such as the discrete dipole approximation (DDA), require a high degree of structural resolution. Such methods are slow, spending considerable time processing redundant (i.e. useless) information. Conversely, when using techniques that incorporate too little structural information, the resultant radiative properties are not physically realistic. Then, we ask the question, what features are most important in determining scattering? This dissertation develops a general technique that can quickly parameterize the important structural aspects that determine the scattering of many diverse snowflake morphologies. A Voronoi bounding neighbor algorithm is first employed to decompose aggregates into well-defined interior and surface regions. The sensitivity of scattering to interior randomization is then examined. The loss of interior structure is found to have a negligible impact on scattering cross sections, and backscatter is lowered by approximately five percent. This establishes that detailed knowledge of interior structure is not necessary when modeling scattering behavior, and it also provides support for using an effective medium approximation to describe the interiors of snow aggregates. The Voronoi diagram-based technique enables the almost trivial determination of the effective density of this medium. A bounding neighbor algorithm is then used to establish a greatly improved approximation of scattering by equivalent spheroids. This algorithm is then used to posit a Voronoi diagram-based definition of effective density approach, which is used in concert with the T-matrix method to determine single-scattering cross sections. The resulting backscatters are found to reasonably match those of the DDA over frequencies from 10.65 to 183.31 GHz and particle sizes from a few hundred micrometers to nine millimeters in length. Integrated error in backscatter versus DDA is found to be within 25% at 94 GHz. Errors in scattering cross-sections and asymmetry parameters are likewise small. The observed cross-sectional errors are much smaller than the differences observed among different particle models. This represents a significant improvement over established techniques, and it demonstrates that the radiative properties of dense aggregate snowflakes may be adequately represented by equal-mass homogeneous spheroids. The present results can be used to supplement retrieval algorithms used by CloudSat, EarthCARE, Galileo, GPM and SWACR radars. The ability to predict the full range of scattering properties is potentially also useful for other particle regimes where a compact particle approximation is applicable.