Improved precipitation forecast accuracy, including precipitation type, location, quantity, time of onset and cessation, is essential for improved transportation safety. One reason why precipitation forecasts are generally poorer than for other forecast variables is that water is the most variable of the major constituents of the atmosphere. It acts over a wide range of temporal and spatial scales and tends to be under observed with conventional upper-air observing systems. This is especially true under conditions of active weather that can significantly impact surface and multi-modal transportation. Aside from the transportation safety issues, including situational awareness and decision support, cost effective maintenance decisions require accurate forecasts and long lead-times to optimize responses to rapidly changing and sometimes dangerous conditions.

Several things have to be accomplished to improve objective and subjective precipitation forecasts for all applications, including surface transportation. Since moisture can be grossly under-observed in time and space, a mixture of conventional and improved surface and upper-air moisture sensing systems have to be developed, tested, and/or deployed. Improved techniques to assimilate direct and indirect moisture observations into high-resolution numerical weather prediction models have to be refined or developed and tested. Methods and tools for real-time verification such as RTVS. Workstations and methods to portray complex data sets and data interrelationships for advanced decision support systems such as MDSS, its components and derivatives must be developed, tested, and put into wide-scale use.

One of the most promising new upper-air moisture sensing systems uses signals from the Global Positioning System (GPS) satellites and collocated surface meteorological sensors to continuously monitor the total quantity of water vapor in the atmosphere. Since 1998, NOAA’s Forecast Systems Laboratory (FSL) and several agencies within the Department of Transportation (DOT) have been evaluating the use NDGPS for improved weather forecasting, climate monitoring, and precise positioning. More recently, FSL has been working with state and local governments, including Florida, Michigan, and Mesa County, Colorado, to incorporate their DGPS sites into the NOAA GPS-Met network. There have already been many benefits from these collaborations, including: expansion of the GPS-Met sensor network at low cost and risk; demonstration of significant improvements in weather forecast accuracy; providing a new and important observation to forecasters and modelers; and densifying the network of surface meteorological sensors across the U.S.

GPS data impact assessments have been carried out at FSL since 1997. Until recently, FSL has used a coarse (60-kilometer) resolution model and a simple data assimilation scheme (Optimal Interpolation) to perform these studies. The impact studies have been made by comparing relative humidity forecasts with twice-daily radiosonde observations at a small number of widely spaced sites. Despite these shortcomings, FSL has demonstrated that the addition of GPS data to weather models: (1) improves the description of the moisture field; (2) that the impact is greatest during active weather; and (3) that the impact is sensitive to the number of stations in the network. While an improvement in the three-dimensional moisture field a necessary precondition for improved precipitation forecasts, it is neither sufficient nor definitive. Tests are needed at higher spatial resolutions using more appropriate assimilation schemes such as variational analysis, and verification on actual precipitation amounts. In addition, the development of case histories will help us understand the complex interactions between the observations and the models.