Assimilation of ZDR Columns for Improving Model Spin-Up

Achieving accurate storm-scale analyses and reducing the spin-up time of modeled convection is an active area of research and is a primary motivation for assimilating radar data. One popular technique for accomplishing this is diabatic initialization, in which latent heat and/or moisture increments are added to induce and sustain updraft circulations. In the current version of the Advanced Regional Prediction System (ARPS) Cloud Analysis, moisture is typically added to locations exceeding an observed Z threshold, while latent heat may be added in regions with positive vertical velocity, often determined via the variational assimilation of radial velocity.

In this study, the ARPS Cloud Analysis is modified to incorporate dual-polarization radar data into the diabatic initialization routine. Using criteria for Z, Z_DR, and ρ_hv, Z_DR columns are detected and, as proxies for updraft location, are used to identify areas of moistening/heating as well as to suppress moistening in areas not associated with Z_DR columns. A set of experiments is performed for the 19 May 2013 tornado outbreak in central Oklahoma using data from KTLX and the Oklahoma Mesonet. Analyses and forecasts from the legacy Cloud Analysis are compared to various implementations of the new Z_DR-based technique. Preliminary findings suggest a marked improvement in analyzed updraft location. Swaths of forecasted 1-6 km updraft helicity show more cohesive, organized tracks with less error in positioning and forward speed compared to observed tornado tracks. Quantitative analysis of Equitable Threat Score for Z also indicates improved performance when using the modified Cloud Analysis routine.