Modeling radar backscattering from melting snowflakes

In our previous study [4], we modeled radar backscattering from simple shapes, like spheres, spheroids and clusters of spheres. We used four different water contents for the coated spheres and the clusters. We used the both DDA and TMM for modeling. The results showed that DDA agreed well with the exact solutions for homogeneous particles, but for water-coated particles, the grid should be large enough in order to preserve skin depth. Using effective-medium approximations in the Rayleigh approximation produced significant errors when compared to the results for the clusters of spheres. It was also noted that using the filtered coupled dipole (FCD) as polarizability in the DDA modeling effectively doubled the accuracy for water particles. We have adopted FCD in our DDA computations. 2. Modeling For the modeling, we use two types of snow particles: an oblate and fluffy, spheroidal shape and a fractal shape. The outer shape is used as a spheroidal surface envelope for the fluffy interior, which is composed of an array of discrete ice dipoles in uniform random positions. To simulate the early melting process, the ice dipoles near the bottom surface of the particle are changed to water. To simulate ground validation, we fit the density of snowflakes in the C-band reflectivity, and use the same density for higher frequencies, which are typical for satellite measurements. In Fig. 1, we show sample spheroidal particles with 20% water content for two different diameters. Notice that the density of the snowflakes depends on the diameter according to [5]. In Fig. 2, we show sample fractal particles with increasing number of iterations. The fractal shapes are generated using the algorithm by [6].