Assimilating IMPACTS Data for 7 Feb 2020 Storm to Improve Winter Cyclone Prediction

Winter cyclones have major impacts on the U.S. Northeast and Midwest every year, including power outages, disruptions of air and ground traffic, injuries, fatalities, and economic losses. The accuracy and lead time of forecasts for these storms is thus critical to minimizing those damages by forewarning the public and emergency managers. This is especially true of forecasts of the heavy precipitation falling from mesoscale snowbands often embedded within these storms. As such, they have long been a focus of study by the meteorological community. We now know many factors favorable for snowband formation, including sufficient moisture, frontogenesis, and conditional symmetric instability. However, there remains much to learn about the inner workings and predictability of both snowbands and winter cyclones. Recently, use of data assimilation, especially with ensemble techniques, have proven to be useful in identifying and reducing initial condition and forecast error. The authors assimilate novel ER-2 and P-3 aircraft observations gathered by the 2020-2023 NASA-funded Investigation of Microphysics and Precipitation for Atlantic Coast-Threatening Snowstorms (IMPACTS) snowbands-focused field campaign to identify changes in feature accuracy as a result of including these unique observations. Doing so should improve model representations of the storm and its snowbands, potentially illuminate more of the dynamics and physics of these cyclones and snowbands, the advantages and disadvantages of assimilating such data, and provide guidance as to how related data can be incorporated to address model deficiencies. These experiments focus on a case study of the 7 February 2020 winter cyclone that produced heavy snow over New York and further Northeast. Observations are assimilated into the PSU-EnKF WRF modeling system, via several experiments (No-DA, conventional obs, and conventional + IMPACTS) with two nested domains focused on the Northeast at 9-km and 3-km horizontal grid spacing. Results will be presented via a combination of methods, including direct comparison with observations, with performance quantified using RMSE and Equitable Threat Scores and neighborhood probability analysis. Further, use of a novel snowband tracking algorithm built using single-linkage clustering of radar reflectivity, combined with ensemble sensitivity analysis via ensemble sub-grouping of storm features, will highlight factors important for or detrimental to the model depiction of the event. The primary focus will be changes in feature location, timing, and predictability, including assessing the duration of effects from assimilation of the field campaign observations, with a focus on radar reflectivity. Preliminary results indicate that assimilating IMPACTS observations at 14-15 UTC produces noticeable changes up to at least 18 UTC, and increases accuracy in the presence and placement of outer snowbands in the northwest edge of the cyclone’s precipitation shield. These results may be useful for future decisions on what observations to pursue and assimilate for refinement of similar features in other storms.