Utilizing Ensemble Sensitivity for Data Denial Experiments on the 4 April 2012 Dallas, Texas Dryline-Initiated Convective Outbreak Using West Texas Mesonet Observations and WRF-DART Data Assimilation

Ensemble sensitivity analysis (ESA) uses the covariance relationships between a chosen forecast aspect and initial condition variables as well as the variance in those initial variables to assess where small initial condition errors can lead to large forecast error. In a convective environment, accurate forecasts of vertical wind shear, convective available potential energy (CAPE), and composite reflectivity (MDBZ) are critical to the issuance of severe weather outlooks. ESA provides the forecaster with an understanding of how observations overlap with regions of large sensitivity in order to make informed decisions about potential errors in the forecast. An additional application of ESA is identifying observation targeting locations where assimilated observations will most reduce the variance of the chosen forecast aspect. Estimations of variance reduction can be made via mathematical relationships of sensitivity, observation variance, and initial condition variance. These estimations can be compared to the actual variance reduction through data denial experiments to understand the usefulness of sensitivity-based observation targeting.

A data denial experiment will be presented that analyzes the variance reduction of MDBZ with respect to withholding West Texas Mesonet (WTM) observations during the 4 April 2012 dryline-initiated convective outbreak over Northern Texas. An eastward surging dryline, along with an advancing cold front, initiated supercell convection west of the Dallas/Ft. Worth metroplex, producing several tornadoes that propagated to the northeast. Using an ensemble Kalman Filter (EnKF) data assimilation technique within the data assimilation research testbed (DART) framework, a 50-member control forecast with 4km grid spacing is produced that does not assimilate the WTM observations. Subsequent forecasts assimilate a varying number of WTM stations and observation types (temperature, dewpoint, sea level pressure) based on observation targeting locations that maximize the expected variance reduction of various convective-based response functions. A summary of expected versus actual variance reduction will be presented, detailing similarities and differences from ESA theory. The fundamental mesoscale predictability of this dryline-initiated convective event will also be briefly discussed.