Model Comparison of the 22-24 January 2016 Nor'easter

The 22-24 January 2016 nor’easter was rated the fourth most severe snowstorm to impact the northeast United States since 1950 and significantly affected the urban corridor, a region home to over 50 million people and of great significance to the U.S. economy.  Numerical weather prediction models correctly predicted snowfall totals exceeding 60 centimeters (~2 feet) over the southern Mid-Atlantic, yet snowfall totals from northern New Jersey and into southern New England were under predicted by 30 centimeters or more.  This study uses a nine-member ensemble of Weather Research and Forecasting (WRF) model simulations, varying only by initialization time (3-5 days prior to the event) or model input source (ECMWF, NAM, or GFS model analysis), to investigate how well these simulations addressed the nor’easter from its initial cyclogenesis through maturity. Particular focus was given to the storm’s northern precipitation edge (specifically, mesoscale snow bands, precipitation type, and radar reflectivity) during the storm’s mature phase when snowfall total varied greatest from observations. Ensemble model output was validated against rawinsondes profiles, NEXRAD WSR-88D radar reflectivity, and ECMWF model analysis.

NAM and GFS-derived nor’easter storm tracks lagged behind and tracked 150 km southward of model analysis due to a 300 hPa height cut off resulting from an excessively strong and prolonged upper-tropospheric right entrance jet region and spatial errors in the location of greatest low-level temperature advection. As a result, the NAM and GFS coastal cyclones tracked too slowly, were too weak, and tracked too far offshore than as indicated from ECMWF model analysis. Of the nine members, only the ECMWF-based WRF run initialized four days prior to the event (Jan. 20) brought snowfall to Long Island which was consistent with ground validation measurements.  Consistent with model analysis and NEXRAD data, this run accurately resolved the location and intensity of the main deformation zone along the Mid-Atlantic coast and also generated a strong, mesoscale snow band propagating from central New Jersey into northern New York City.  Even in this best WRF forecast, snowfall did not reach as far northward or reach similar depths as indicated from analysis data from New Jersey through southern New England which likely stems from errors with both simulated model dynamics and microphysics during late-stage nor’easter events.