Events / Media

Data Assimilation Seminar Series

Prof. Steven J. Greybush (Mar. 6, 2017, 16:45-)

Affiliation Department of Meteorology and Atmospheric Science, The Pennsylvania State University
Title Ensembles, Data Assimilation, and Predictability for Winter Storms
Abstract Ensemble data assimilation and prediction systems quantify uncertainties with flow-dependent, probabilistic forecasts. Careful evaluation of ensemble system design and performance are needed to provide reliable ensemble spread, identify sources of forecast error, and characterize the predictability for extreme weather events.

The ensemble predictability of high impact east coast winter storms of January 2015 and 2016 are assessed. Skill scores and reliability diagrams indicate that the large ensemble spread produced by operational forecasts was warranted given the actual forecast errors imposed by practical predictability limits. Predictability horizon diagrams depict the forecast lead time in terms of initial detection, emergence of a signal, and convergence of solutions for an event. For the 2015 storm, uncertainties along the western edge's sharp precipitation gradient are linked to position errors of the coastal low, which are traced to positioning of the preceding 500 mb wave pattern using the ensemble sensitivity technique. Comparisons of regional convective-scale ensemble simulations provide insights to the intrinsic and practical predictability of these events.

Next, the relative contributions of synoptic scale, mesoscale, and model errors to the spread and skill of forecasts for intense mesoscale lake-effect snow (LES) events are investigated using a regional convection-allowing data assimilation and ensemble prediction system with varying initial and boundary conditions, methods of perturbations, and physical parameterizations. Performance in simulating a long lived LES event sampled by the OWLeS field campaign in December 2013 is evaluated, including the use of object-based approaches to quantify snowband morphology. Synoptic-scale perturbations (via model lateral boundary conditions) play an influential role in LES precipitation forecasts at longer lead times of several days, whereas the improvement in initial conditions from a regional ensemble data assimilation is realized at shorter lead times. These uncertainties impact the timing and intensity of predicted precipitation, as well as band location and orientation, giving insight into the timescales of practical predictability of LES.

Finally, the potential for assimilation of dense networks of smartphone surface pressure observations to improve convective-scale ensemble forecasts is explored.