MPEX: Improving forecasts of severe convective storms through targeted observations

Forecasts of deep convection remain a considerable challenge, but recent progress in convection-permitting ensemble forecast methods provide an opportunity to improve guidance. This progresses hinges on defining an appropriate ensemble initial state that represents initial condition (IC) uncertainty, and evolving this uncertainty in the ensemble forecast to map how forecast outcomes are sensitive to this IC uncertainty. From this, locations for targeted observations can be identified that have the potential to improve the reliability of convective forecasts.

During Spring 2013, the Mesoscale Predictability Experiment (MPEX) sought to improve short-term (here 6-15 h) forecasts of convective weather episodes over the Great Plains. MPEX approached this challenge by increasing observation density with mesoscale spaced (~150 km) dropsonde observations in the vicinity of disturbances upstream of uncertain convective weather events. Potentially uncertain ICs were identified in part by formal ensemble sensitivity analysis (ESA) from longer-range forecasts. Fifteen dropsonde missions were conducted, which included events with a wide range of convective organization.

In retrospective case studies, the impact of the additional dropsonde observations is determined through a set of data denial experiments with the WRF/DART ensemble data assimilation system and subsequent ensemble forecasts. Verification of ensemble forecasts is made against conventional and research observations. Further, we assess the relative utility of ESA to identify a priori targeted observation locations for convection forecasts. At the conference, a brief overview of MPEX will be presented along with a discussion of results.