Forecast Performance of an Operational Mesoscale Modeling System for Post- Tropical Storm Sandy in the New York City Metropolitan Region

On 28-31 October 2012, Post-Tropical Cyclone Sandy severely impacted the northeastern portion of the United States, including New Jersey, Connecticut and the New York City metropolitan area, four days after being initially classified as a Category 1 Hurricane. During the period between first attaining hurricane strength south of Jamaica on October 24th and its landfall near Brigantine, New Jersey on October 28th, Sandy weakened and strengthened several times. In preparation for the expected landfall along the central New Jersey coast, many municipalities as well as private and public sector agencies deployed resources based upon anticipated hurricane intensity, although Sandy had weakened to a tropical storm and then became a post-tropical cyclone just prior to landfall. Although. the storm produced widespread heavy rainfall, the most significant impacts were as a result of storm surge along the coast as well as high winds inland. Wind gusts ranged from 70 to higher than 90 miles per hour both along the coast and inland. The storm's large size with tropical storm force winds extending nearly 500 miles from the center prior to landfall resulted in extensive coastal flooding and infrastructure damage from the storm surge and an extended period of strong winds. Widespread power outages resulted in over eight million people affected; some were without power for more than two weeks.

In our continuing work focused on providing weather-sensitive business solutions, IBM's “Deep Thunder” service provides operational forecasts twice daily for areas of southeastern New York State and northern New Jersey. With an operational history that spans more than a decade, producing one to three day model-based forecasts at one to two kilometer resolution, the overall model configuration has evolved and improved over time to reflect improvements in NWP model capability, availability of relevant input data sets as well as computational efficiency. Over the past several years, the system has focused on producing 84-hour predictions updated every 12 hours. The NWP model configuration that was operational during the time Sandy impacted the northeastern United States was based upon the WRF-ARW community model. This configuration, which has been in operation since early 2009, is a nested configuration, with the highest resolution at two km, utilizing 42 vertical levels. The specific configuration that was operational in October of 2012 was based upon version 3.31 of WRF-ARW and also included parameterization and selection of physics options appropriate for the range of geography within the model domain from highly urbanized to rural. This included WSM-6 microphysics (explicit ice, snow and graupel), Yonsei University non-local-K scheme with explicit entrainment layer and parabolic K profile in the unstable mixed layer for the planetary boundary layer (PBL), NOAH land-surface modelling with soil temperature and moisture in four layers, fractional snow cover and frozen soil physics, Grell-Devenyi ensemble cumulus parameterization, and the 3-category urban canopy model with surface effects for roofs, walls, and streets. Background surface data included high resolution surface terrain derived from NASA 30m SRTM and SPoRT 1km resolution SSTs.

Given the model length and frequency, the system produced ten operational forecasts that covered the period prior to landfall and the impact in New York and New Jersey. The system exhibited proficient skill in forecasting regional as well as local scale impacts of Sandy with significant lead time. In particular, with the model run initialized at 12 UTC on 27 October 2012 and those produced afterwards, the model forecasted local scale impacts of Sandy as it made landfall as a post-tropical storm.

In order to evaluate the relative skill improvement and quality of the forecasts produced by Deep Thunder at a storm-scale, we compare the model results of forecasts made with the aforementioned model configuration, which were generated operationally prior to Sandy's landfall, with an updated configuration of the model run as a “hindcast”, which utilizes the WRF-ARW version 3.4.1 model as well as additional features. These features include an extension of a more advanced implementation of three-dimensional variational data assimilation of observations to several thousand surface and near-surface observation stations operated by Earth Networks, NOAA and other agencies. The representation of the surface conditions was improved through the use of additional remotely sensed observations provided by NASA and USGS (e.g., SRTM terrain data, SPoRT SSTs, land use, vegetation, soil), and the new Thomspon double-moment micro-physics scheme, as well as the Yonsei PBL scheme with topographic wind correction and the Grell-3D cumulus parameterization. Both model configurations utilized the NCEP NAM for setting background fields and lateral boundaries.

The results of both model configurations were compared with observational data and other available forecasts as well as the operational availability of specific forecast products. Such performance is examined by considering forecast timing, locality, and intensity of the storm impacts as well as through the utilization of traditional and spatial verification methodologies.