The utility of Ensemble Kalman Filter-based modeling for short-term forecasts of banded snow

Numerical forecasts of a banded precipitation event using the Ensemble Kalman Filter (EKF12) are compared to forecasts similar to those used operationally. These include a 12- and 3-km grid-spaced deterministic forecast (D12 and D3, respectively) and a 30-member pure ensemble forecast (E12). Numerical model output is compared to available observations and the North American Mesoscale (NAM) model analyses to see if EKF12 provides superior guidance. The D12 and D3 experiments have some of the so-called ingredients of banding, but lack the upstream instability observed during this event and do not produce banding. Nevertheless, both experiments have the precipitation in the same location as the observations. The E12 ensemble mean forecast has a favorable, albeit weak, configuration of the ingredients and has the precipitation in a similar location to the observations, but individual members show a high spread with several exhibiting flow patterns not favorable for banding and with no bands. The EKF12 ensemble mean shows relatively good agreement with conceptual models of banding, the magnitudes of the ingredients are close to those in the NAM analysis, and the ensemble mean even reproduces a double-banded structure similar to the observations but it places the ingredients and bands about 100 km too far north.