An observations-based, statistical system for short-term probabilistic forecasts of aviation-sensitive weather parameters

Accurate and timely short-term (< 6 h) forecasts of high-impact aviation weather parameters, such as low ceiling, poor visibility, and thunderstorms are a critical component to airline operations. Adverse weather conditions not only plague the efficiency of commercial operations, but are also a safety concern, particularly for General Aviation pilots and passengers. A review of weather-related Accident Briefs indicates that the vast majority of fatal accidents (over 80%) is caused by low ceiling, poor visibility, and thunderstorms. Moreover, the number of aircraft delays has increased markedly – up over 60% in September 1999 from a year earlier.

Clearly, improved weather guidance would be of great value to the aviation industry, especially if the guidance provided a measure of the uncertainty associated with the forecasts. Traffic-flow-management personnel recognize the inherent uncertainty when predicting weather, and attempt to account for it through careful cost-benefit decision-making – all in an effort to minimize the airlines’ operating costs. The uncertainties incorporated into cost-benefit analyses can best be captured, then, via reliable probabilistic forecast guidance. This suggests that short-term, observations-based, statistical forecast techniques that yield a quantitative measure of uncertainty would be most useful to the aviation industry, especially if such techniques would provide sharp, timely, and reliable probabilistic forecasts of high-impact aviation weather parameters.

A framework for an automated system that provides probabilistic guidance for aviation weather parameters is presented. Preliminary results from early versions of the automated system are also presented. The system utilizes NEXRAD Information Dissemination Service (NIDS) radar data, surface mesonet data, and Velocity Azimuth Display (VAD) wind information as input to a stepwise regression procedure. Output consists of probabilistic categorical forecasts for lead times of 6, 12, 18, 30, and 60 min. The system is capable of providing updates every 6 min, thereby accounting for rapid changes in local conditions. Forecasts are made for a selected airport and for an array of grid points centered on the selected airport. The array covers a domain comparable to the approach-control zone around the selected airport (i.e., the area within a radius of about 60 km of the airport). Thus, the system is designed to provide probabilistic categorical forecasts of aviation-sensitive weather parameters at the airport as well as the spatial distribution of probabilities of certain weather categories (such as the presence of thunderstorms) in the region surrounding the airport. In this manner, air-traffic controllers can obtain high-resolution, rapidly-updated guidance for improving air-traffic flow.

It is expected that this observations-based system will provide an advancement in two aspects of aviation: 1) reliable, ultra-short-term probabilistic forecasts will ameliorate air-traffic congestion during adverse conditions, thereby reducing airline costs and passenger complaints, and 2) short-term guidance on dangerous weather elements in the near vicinity of airports can diminish the risk of General Aviation accidents.