The Influence of Assimilating Targeted Observations upon Ensemble Forecasts of Convection Initiation

In this study, we evaluate the influence of assimilating targeted meso-α- to synoptic-scale observations collected in the upstream, pre-convective environment upon subsequent short-range (3-12 h) ensemble forecasts of convection initiation across the central United States for the fifteen aircraft missions conducted by the Mesoscale Predictability Experiment (MPEX) in May and June 2013. Utilizing the ensemble Kalman filter implementation within the Data Assimilation Research Testbed software package as coupled to version 3.4.1 of the Advanced Research version of the Weather Research and Forecasting mesoscale model, two nearly-identical thirty-member ensembles of short-range forecasts are conducted for each mission. In one, initial conditions are generated through a cycled data assimilation process that incorporates the targeted MPEX dropsonde observations from that day's mission. In the other, initial conditions are generated through a cycled data assimilation process that excludes the targeted MPEX dropsonde observations. All forecasts for a given mission begin at 1500 UTC, extend forward 15 h, and are conducted on a domain encompassing the conterminous United States with 3 km horizontal grid spacing and 51 vertical levels.

Convection initiation events across the central United States are identified from objects in the convectively-active field with observed or simulated radar reflectivity of 35 dBz or greater at the -10 C level for at least thirty minutes. All convection initiation events occurring between 1500 UTC and 0600 UTC are considered, with particular focus given to initial convection initiation events, or those occurring in convectively-undisturbed environments (e.g., no active convection within the last 3 h and 250 km). Verification is conducted over multiple spatiotemporal thresholds: within 40 km/1 h, 80 km/1.5 h, 120 km/2 h, 160 km/2.5 h, and 200 km/3 h of an observed initial convection initiation event. It is found that the Brier skill score, a widespread metric by which the skill of probabilistic forecasts may be assessed, is ill-suited for assessing the skill of probabilistic forecasts of rare events at each of these thresholds. Furthermore, both ensembles are found to overpredict the number of total convection initiation events by a factor of approximately two for each of the fifteen missions considered. The presentation will discuss alternative means of assessing the skill of probabilistic forecasts of rare events, as applied to simulation output from each MPEX missions, with the goal of quantifying the influence that assimilating targeted observations has upon forecast skill for the events considered.