High-Resolution Snowstorm Measurements and Retrievals Using Cross-Platform Multi-Frequency and Polarimetric Radars

Many studies have performed microphysical retrievals using radars of different frequencies, platforms, and methodologies. However, little is known about the consistency of retrievals derived from different radar platforms (i.e., airborne or spaceborne vs. ground-based) and their various methodologies. This presentation will show the first ever results that directly compare multiple snow mass-weighted mean diameter (D_m) retrievals from both nadir-pointing airborne multi-frequency radars and ground-based polarimetric range-height indicator (RHI) radar scans taken along the same airborne flight track. The ability to directly collocate RHI scans with overpassing airborne radar data produces much higher spatial [O(10 m)] and temporal [O(1 s)] resolution than other inter-platform comparison efforts which are often restricted to O(1 km) and O(100 m) spatial and O(1-10 min) resolution. The radar data used for this comparison is from the February 25, 2022 NASA-led Investigation of Microphysics and Precipitation for Atlantic Coast-Threatening Snowstorms (IMPACTS) field campaign case in Albany, New York. Radar measurements from these two separate platforms were interpolated onto a common 50 m x 50 m range and height grid. Figure 1 shows three of the eight total overpass periods with collocated radar measurements from both the Albany-stationed RaXPol mobile ground radar and the ER-2 aircraft. As shown in Figure 1, both ER-2 and RaXPol radars show incredibly detailed and consistent sub-kilometer X-band reflectivity features (second and third rows). The single wavelength and dual wavelength ratio (DWR) radar data taken from these overpass periods were then used to retrieve D_m (see Figure 2) using three of the most recently devised empirical and semi-empirical methodologies: An empirical X-W band third order polynomial relation (DWR_X-W Poly), a Ku-Ka band Neural Network model (DWR_Ku-Ka NN), and a dual-polarization bivariate power-law relation using reflectivity and specific differential phase (RaXPol). D_m statistics between each retrieval method show the largest mean absolute errors (MAE) between RaXPol and DWR_Ku-Ka NN (MAE=0.92 mm) whereas the lowest MAE was between the two ER-2 DWR-based retrievals (MAE=0.49 mm). RaXPol D_m retrievals were generally the largest (with biases of 0.44 mm and 0.86 mm between each other retrieval) whereas DWR_Ku-Ka NN was generally the smallest (with biases of 0.42 mm and 0.86 mm between each other retrieval). In addition to the radar and retrieval comparison, we also performed a triple-frequency radar diagram analysis using the ER-2 aircraft radars to investigate the microphysical characteristics of the snow particles in Albany throughout the IOP. Discrete dipole approximation scattering simulation results plotted alongside these triple frequency diagrams suggest that snow aggregates were generally composed of needles. These results can be used as a benchmark for comparing retrieval methodologies and highlight the continued uncertainty regarding the optimal approach for ice microphysical retrievals.