An Investigation of the Conditional Practical Predictability of the 31 May 2013 Heavy-Rain-Producing Mesoscale Convective System

Abstract

On 31 May 2013, strong thunderstorms initiated in west-central Oklahoma with one of the storms eventually creating a very strong tornado near El Reno, OK. The storms then grew upscale into a quasi-stationary mesoscale convective system that produced prolonged heavy rainfall that led to severe flooding across parts of Oklahoma, including the Oklahoma City metropolitan area. A 50-member ensemble of short range (0-24 h) forecasts was conducting using a set of initial conditions generated via cycled data assimilation to quantify event predictability and identify forecast sensitivities, primarily with CI and initial upscale growth. Both a composite and ensemble sensitivity analysis were performed using fractions skill score (FSS) as the partitioning forecast metric to measure these dependencies. Both analyses indicate that better forecasts are associated with an upper-level trough further to the west, stronger upper-level ridging to the east, stronger low-level convergence across much of Oklahoma, and a further west dryline and quasi-stationary front across northwest Oklahoma and Kansas. These conditions were associated with better early-stage MCS forecasts, which appears to be associated with better CI forecasts. These findings suggest that even slight differences in conditions leading to CI had large sensitivities in CI and the subsequent upscale growth.