In our continuing work focused on developing advanced model applications to enable weather sensitive business solutions, IBM's Deep ThunderTM
service provides operational forecasts twice daily for the New York City metropolitan region including areas of southeastern New York State and northern New Jersey. With an operational history that spans almost two decades producing 1 to 3.5 day model forecasts at one- to two-kilometer resolution, the overall model configuration has evolved and improved over time to reflect improvements in Numerical Weather Prediction (NWP) capability as well as computational efficiency. Over the past several years the system has focused on producing 84-hour predictions updated every 12 hours, which included applications associated with the operations of the urban infrastructure. The NWP component was derived from the WRF-ARW community model. It operated in a nested configuration, with the highest resolution at two kilometers, utilizing 51 vertical levels. It used parameterization and a selection of physics options appropriate for the range of geography within the domain from highly urbanized to rural. This included: RRTMG short wave and long wave radiation schemes, Thompson microphysics (explicit ice, snow and graupel), MYNN2 for the planetary boundary layer, NOAH land-surface modelling with soil temperature and moisture in four layers, fractional snow cover and frozen soil physics and Grell-Freitas cumulus parameterization for the outer nest. A number of background surface fields were derived from NASA data sets including model orography, sea surface temperature, land use and green vegetation fraction. NCEP RAP 13km analysis was used for background fields and hourly output of the NCEP GFS was used for lateral boundaries.
We will discuss the implementation of an updated implementation of Deep Thunder, which utilizes a four-nested configuration down to 667m and larger domain size of 100 x 100 grid points. The updated model framework and configuration is intended to serve as an operational, research, development and collaboration urban testbed for both business applications and the NWP performance at this scale. The testbed will provide the basis for exploring advanced model configurations including larger domains, more sophisticated urban and physical configurations including the NOAH-MP land surface model, urban land use and multi-layer urban canopy models as well as additional data sources and assimilation techniques to support coupled model applications for impact studies. The model testbed is intended to serve as a research to operations platform, which enables the exploration, and development of new techniques and applications within a production environment.