"This award is funded under the American Recovery and Reinvestment Act of 2009 (Public Law 111-5)."
Severe thunderstorms and associated local extreme weather events (e.g., tornadoes, floods and lightning) significantly threaten life and property but remain difficult to forecast with precision. If sufficiently reliable, detailed forecasts and advance warnings of these threats could trigger defensive actions and mitigate resultant losses. However, at lead times sufficient to allow adequate preparations by those in the path of an intense developing storm (or more particularly, a storm that has yet to develop), deterministic forecasts currently in broad use are generally not accurate enough to provide useful guidance, and may even miss critical events altogether. By contrast, probabilistic approaches that rely upon ensembles of multiple computer-based forecast model runs, offer a promising alternative, but such output must be properly interpreted to have optimal value.
This investigative team will develop methods to optimize probabilistic forecasts by identifying the smallest required changes of initial atmospheric conditions (which are in-turn used to set computer-based weather simulations into motion) that consequently lead to the most extreme -simulated severe weather outcomes. Solving this problem also provides a measure of the probability of a given severe weather event. Forecasts so derived are termed "exigent" because of the tandem need for extreme precision in their formulation and urgent lead-time requirements to allow lifesaving societal responses. This solution will be pursued through application of modified forms of a technique termed four-dimensional variational data assimilation (4d-VAR) techniques, which make optimal use of observations collected in the environment encompassing developing storms, as developed in conjunction with the Weather Research and Forecasting Model (WRF). The short-term technical objective of this study is to demonstrate successful exigent forecasting of severe thunderstorm precursors and thereby characterize the probability of worst-case scenarios on a location- and time-specific basis. Investigators will apply these techniques to identify conditions that maximize the potential for tornadogenesis and delineate these probabilities as a function of time and place via evolving county-scale maps of a maximum significant tornado parameter (STP). The STP has been subjectively shown to discriminate convective events that involve discrete supercells and produce disproportionally large, damaging tornadoes.
The intellectual merit of this approach is embodied by a new approach to refinement of ensemble forecasting techniques that could overcome limitations of current deterministic forecast methods, one which might have applications as diverse as improved forecasts for landfalling tropical cyclones and wintertime cold-air outbreaks impacting citrus crops. Broader impacts could ultimately include reduced loss of life and property, and will more immediately support training of graduate and undergraduate students assisting with this research.
