While lake-effect snows are a common sight around the wintertime Great Lakes, challenges remain in providing accurate forecasts of the timing, location, and amounts of precipitation. This work seeks to quantify the predictability of these meso-gamma-scale storms using a regional convection-allowing data assimilation system, and the ability to which these estimates can be improved with a strong emphasis on operations. The numerical experiments highlight an event which occurred and was sampled during the OWLeS field campaign of 2013-2014, which provided additional sources of verification data. Coupling WRF with the Penn State EnKF, strong sensitivities were found to choices of parameterization schemes, lake surface temperature input, and moreover, initial and lateral boundary conditions. Ensembles initialized with perturbations created by the NMC method do not retain enough spread to be useful, collapsing even with increased relaxation and multiplicative inflation. An ensemble initialized using individual GEFS members for both initial and boundary conditions maintains spread and provides a useful forecast. Further analysis shows that assimilation has a significant and positive impact on the simulations as compared to observations and precipitation products. Additionally, observation influence appears to extend up to 24 hours past the final analysis period. Overall, this ensemble design shows promise for enhancing short- and long-term prediction of these extremely localized but high-impact events.

---

*email: sfs153@psu.edu
*Preference: Poster