Evaluation and Utilization of Meso-γ-Scale Numerical Weather Prediction for Logistical and Transportation Applications

The effectiveness and efficiency of logistical operations in a number of industries are dependent on the response of local transportation systems (surface and air) to weather conditions. To enable proactive operations, planning and scheduling need to be driven by predictions of weather events that can impact them. Further, those predictions need to be at a temporal and spatial scale consistent with such activities. Consider, for example, the support and execution of operations in and around a major metropolitan area with a number of transportation facilities. Hence, meso-γ-scale numerical weather models operating at higher resolution in space and time with more detailed physics may offer the appropriate precision and accuracy within a limited geographic region for these problems.

To begin to explore the relevance of this idea, we build upon the earlier efforts by IBM Research to implement an operational testbed, dubbed "Deep Thunder", which has been customized for transportation applications. The original prototype provides nested 24-hour forecasts for the New York City metropolitan area to 1 km resolution utilizing explicit, bulk cloud microphysics. It was extended in 2004 to provide forecasts for other metropolitan areas. In particular, this included model runs covering the greater Baltimore and Washington metropolitan areas at 2 km resolution. It is imbedded in a region at 8 km covering all of Maryland and Delaware, and most of Pennsylvania, Virginia and New Jersey. The system was set up to generate a number of customized visualizations viewed via a web browser, which are updated typically twice per day. All of the prerequisite processing is completed in less than an hour on relatively modest hardware to enable timely dissemination of model results at reasonable cost. These forecast products were accessed by USAF Weather Operations staff based in the aforementioned geographic area for several months during 2005, starting in mid-winter. We will discuss how those products were used and their quality with respect to specific weather events as well as other available forecasts. We will also present some results concerning the overall effectiveness of such modelling capabilities and this particular approach for these applications and recommendations for future work.