Evaluation of Eta model forecasts as a backup weather source for CTAS

Abstract

Knowledge of present and future winds and temperature is important for air traffic operations in general, but is crucial for Decision Support Tools (DSTs) that rely heavily on accurately predicting trajectories of aircraft. One such tool is the CenterTRACON Automation System (CTAS) developed by NASA Ames Research Center. The Rapid Update Cycle (RUC) system is presently the principal source of weather information for CTAS.' RUC provides weather updates on an hourly basis on a nationwide grid with horizontal resolution of 40 km and vertical resolution of 25 mb in pressure. However, a recent study of RUC data availability showed that the NWS and NOAA servers are subject to frequent service interruptions. Over a 210 day period (4/19/0011/11/00), the availability of two NOAA and one NWS RUC server was monitored automatically. It was found that 60 days (29%) had periods of one hour or more where at least one server was out, with the longest outage lasting 13 hours on 9/21/00. In addition, there were 9 days (4%) for which all three servers were simultaneously unavailable, with the longest outage lasting 6 hours on 5/7/00. Moreover, even longer outages have been experienced with the RUC servers over the past several years. RUC forecasts are provided for up to 12 hours, but these are not currently used in CTAS as back up sources (except that the 1 or 2 hour forecasts are used for the current winds to compensate for transmission delays in obtaining the RUC data). Since RUC outages have been experienced for longer than 12 hours, it is therefore necessary to back RUC up with another weather source providing long-range forecasts. This paper examines the use of the Eta model forecasts as a back-up weather source for CTAS. A specific *This work was performed for the Federal Aviation Administration under Air Force Contract No. F19628-00-C-0002. tCopyright © 2001 by M.I.T. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission. output of the Eta 32 km model, namely Grid 104, was selected for evaluation because its horizontal and vertical resolution, spatial extent and output parameters match most closely those of RUC. While RUC forecasts for a maximum of 12 hours into the future, Eta does so for up to 60 hours. In the event that a RUC outage would occur, Eta data could be substituted. If Eta data also became unavailable, the last issued forecasts could allow CTAS to continue to function properly for up to 60 hours. The approach used for evaluating the suitability of the Eta model and RUC forecasts was to compare them with the RUC analysis output or 0 hour forecast file, at the forecast time. Not surprisingly, it was found that the RUC model forecasts had lower wind magnitude errors out to 12 hours (the limit of the RUC forecasts) than the Eta model had. However, the wind magnitude error for the Eta model grew only from 9 ft/s at 12 hours (comparable with RUC) to 11 ft/s at 48 hours. We therefore conclude that RUC forecasts should be used for outages up to 12 hours and Eta model forecasts should be used for outages up to 60 hours.