Evaluating Data Assimilation and Parameter Estimation on the Prediction of Heavy Snowfall during East Coast Winter Storms

During the wintertime, strong surface cyclones can pose a serious threat to life and property for millions of people. From near-zero visibility in blizzard conditions to significant ice accumulations causing widespread power outages, these winter storms can severely disrupt daily life, with impacts often persisting for several days following the event. Accurate prediction of the strength, timing, and location of these storms is critical for minimizing these effects; however, they prove particularly challenging to forecast, in part due to the complex microphysical processes occurring within the storm. Numerical weather models implement parameterization to represent these processes on a finite grid resolution. Some physics parameters can additionally be modified to influence model output; however, an ideal combination of their values is difficult to determine and can vary between events. One possible approach to determining parameter combinations for a particular winter storm is to apply data assimilation (DA) techniques to estimate their values using available observations. This study aims to conduct parameter estimation on winter storm cases featured in the Investigation of Microphysics and Precipitation for Atlantic Coast-Threatening Snowstorms (IMPACTS) using the Weather Research and Forecasting (WRF) model. Between one and three parameters pertaining to cloud microphysics will be selected for estimation based on thorough sensitivity testing to determine any detectable effects on modeled cloud particles and total precipitation. This study will also present an analysis of these sensitivity tests to improve the understanding of how numerical models resolve potential impacts from winter weather events. Many of these storms contained especially impactful regions of localized heavy snowfall; as such, the resolved location and intensity of these “snow bands” will be targeted as the response variable for quantifying the effects of this parameter estimation approach. This procedure will be evaluated against data assimilation techniques that solely involve state estimation to determine any potential future applications in East Coast winter weather forecasting.