Studies have indicated that both initial condition and model errors within atmospheric models can propagate upscale as well as downscale, which limits the range of practical predictability in numerical forecasts (Hohenegger and Schär 2012). With short-term storm-scale prediction to play an increasingly important role in tornado warning operations (e.g., Warn-on-Forecast), it is beneficial to understand the relative impact of errors from the background mesoscale environment on fine-scale features associated with supercell tornadogenesis.
To investigate the effect of relatively small mesoscale errors on submesocyclone-scale vortex development in simulated supercells, perturbations are randomly drawn from typical 1-hour forecast errors observed from the 13 km RUC model and applied to the 29 May 2004 Geary, OK sounding using the method of Cintineo and Stensrud (2013). Three sounding ensembles are created by scaling these errors to 10%, 25%, and 50% of their original magnitude. These are then used to initialize horizontally homogeneous environments for three sets of 20 idealized simulations at a horizontal resolution of 100 m (plus a control forecast). The Vortex Detection and Classification (VDAC) algorithm outlined in Potvin (2013) was used to identify vertically continuous submesocyclone-scale vortices. Finally, a case study was conducted between six members of the 50% perturbation ensemble to examine specific impacts of environmental conditions on processes related to vortex development.
Statistical analysis of vortices detected by the algorithm showed that the distribution of total detections per ensemble member were not statistically different in the 10% and 25% ensembles, suggesting that there may be a threshold in error reduction beyond which improved analysis of the background mesoscale state in which the storm forms will not improve forecasts. However, reducing the initial state errors may improve the prediction of the spatial “envelope” in which the vortices occur. The spread of vortex locations perpendicular to the storm path as well as the length of time over which vortices develop decrease as the initial error magnitude decreases. Distributions of observed vortex tangential velocity and radius differed between ensembles without a clear pattern, although ensemble maxima of these characteristics appear to be correlated with the LCL, LFC, CAPE, and CIN of the initial environment. The results of the case study suggest that location and intensity of cold pools near the rear flank, which appear to be correlated to a slight reduction in low-level moisture, may be responsible for significant changes in overall vortex statistics in certain realizations.
Brittany Dahl (2014). Sensitivity of Vortex Production to Small Environmental Perturbations in High-Resolution Supercell Simulations School of Meteorology, University of Oklahoma.
The data for the thesis can be released on request. It is sufficiently large (2.2TB per simulation) that it is difficult to put online easily but we are glad to share if you can make use of the data and if you can receive the data somehow. It is currently stored at NICS.
Created by amcgovern [at] ou.edu.
Last modified June 12, 2017 12:57 PM