Publications

An important part of the FAA Aviation Weather Development Program is a system, the Integrated Terminal Weather System (ITWS), that will acquire data from the various FAA and National Weather Service (NWS) sensors and combine these with products from other systems (e.g., NWS Weather Forecast Offices and the FAA Aviation Weather Products Generator). This wide variety of input data and products will enable the ITWS to provide a unified set of weather products for safety and planning/capacity improvement for use in the terminal area by pilots, controllers, terminal area traffic managers, airlines, airports, and terminal automation systems (e.g., Terminal Air Traffic Control Automation (TATCA) Center Tracon Advisory System (CTAS) [Andrews and Welch, 1989] and wake vortex advisory systems.

Aviation operations in the airport terminal area, where flights converge from a number of directions onto one or two active runways, create a fundamental limitation on the capacity of the national airspace system. The U.S. Federal Aviation Administration (FAA) has recognized that the throughput of existing terminals can be increased significantly by providing the terminal air traffic control team with Terminal Air Traffic Control Automation (TATCA) tools that increase the efficiency of individual controller tasks and provide a dynamic, overall plan for traffic management throughout the terminal control region (Andrews and Welch, 1989). This latter function relies on accurate projection of traffic flow into the future (0-30 minutes) in order to automatically examine the many possible permutations of control actions. The result is a coordinated plan for the multiple (four to ten) control positions involved in the decision making processes that determine end-capacity at the runways. The FAA has launched an intensive effort to develop and implement TATCA capabilities by taking advantage of preparatory work done at NASA Ames Research Center, MITRE Corporation, and M.I.T. Lincoln Laboratory. An initial TATCA configuration, the Final Approach Spacing Tool (FAST), will be evaluated in the field beginning in 1993 and will be scheduled for possible national implementation two years later. Estimates of the economic value of TATCA-generated operational improvements, when implemented at major airports nationwide, are expected to be over $1 billion yearly by the year 2000 in reduced fuel consumption, other air carrier operating costs, and passenger time (Boswell et al., 1990). Since TATCA is first and foremost a planning system, the primary impacts of weather upon T ATCA performance involve disruption of planning. This can occur because of sudden or unexpected changes in routing, runway availability, or separation standards. In addition, errors in estimated wind produce errors in time-to-fly predictions made by the TATCA planning logic. The TATCA system must be robust with respect to weather events that commonly occur in its region of operation. This paper describes an initial study of the weather information requirements for TATCA, and their relationship to current and future systems for measurement, integration, forecasting and dissemination of meteorological data in the terminal area. A major goal is to stress the need for close coupling between ongoing initiatives in weather sensing/forecasting in the airport terminal area, and air-space capacity enhancement programs.

The characteristics of microbursts and gust fronts, two forms of aviation-hazardous low altitude wind shear, are presented. Data were collected with a prototype terminal Doppler weather radar and a network of surface weather stations in Memphis, Huntsville, Denver, Kansas City, and Orlando. Regional differences and features that could be exploited in detection systems such as the associated reflectivity, surface wind shear, and temperature change are emphasized.

The Federal Aviation Administration (FAA) currently uses the anemometer-based Low Level Wind Shear Alert System (LLWAS) as the primary method of wind shear detection at major U.S. airports. With the upcoming deployment of the Terminal Doppler Weather Radar (TDWR) system, potential methods for integrating the two systems are being investigated. By integrating, advantages of both sensor systems can be utilized. Advantages of the LLWAS ground sensor network include true wind direction measurements, a high measurement frequency, a lack of sensitivity to clear air reflectivity, and few false alarms from radar point targets such as planes, birds, etc. Advantages of the radar include complete scan coverage of the region of concern, the ability to predict events, fewer terrain problems such as sheltering which can reduce the wind speed readings, and almost no false alarms due to non-hazardous wind shear such as thermals. The objectives of this study are to gain a clearer understanding of the basic relationship between the wind information provided by these two very different sensing systems, and to determine the impact this relationship may have on integration of the two operational systems. A proposed mathematical technique for "correcting" LLWAS winds where needed to better match radar winds is evaluated for cases of microburst (divergent) and gust front (convergent) wind shear.

The observability of microbursts with single-Doppler radar is investigated through comparison of radar data and surface weather sensor data. The data were collected during 1988 in Denver, CO as part of the FAA Terminal Doppler Weather Radar measurement program. Radar data were collected by both and S-band and C-band radar, while surface data were taken from a mesoscale network of 42 weather sensors in the vicinity of Denver's Stapleton International Airport. Results are compared with previous similar studies of observability using data from 1987 in Denver, and 1986 in Huntsville, AL. A total of 184 microbursts impacting the surface mesonet were identified. For those microbursts for which both radar and surface data were available, 97% were observable by single-Doppler radar. This compares to 94% observability during 1987 in Denver, and 98% during 1986 in Huntsville. Two strong microbursts (at lease 20 m/s differential velocity) were unobservable by radar throughout their lifetime: one due to low signal-to-noise ratio, and the other due initially to an asymmetric outflow with low signal-to-noise ratio also a contributing factor. Two other microbursts, with differential velocities from 10-19 m/s, were unobservable by radar: one due to shallow outflow with a depth limited to a height below that of the radar beam, and one due to asymmetric outflow oriented unfavorably with respect to the radar viewing angle. Consistent with previous observations, microburst occurrence was most frequent during June and July, when 94 microbursts were identified on 20 days. An anomalously high frequency was also seen in April, although the strength of these events was relatively modest. As expected, the diurnal distribution shows the late afternoon to be the most favorable time for microburst development; more than half of all events reached their maximum strength between the hours of 2-5 p.m. local time.

As part of the FAA Terminal Weather Doppler Weather Radar (TDWR) measurement program in Huntsville, AL and Denver, CO during 1986 and 1987, respectively, the ability of a single Doppler weather radar to observe microburst outflow signatures (i.e., show identifiable radial velocity patterns) was assessed by comparing radar-observed microbursts with those identified by joint use of both radar and data from a mesoscale network (mesonet) of surface meteorological stations (Clark, 1988; DiStefano, 1988). Observability by radar must be considered together with pattern recognition algorithm performance for observable microbursts (Campbell et al., 1988) in order to fully assess the potential effectiveness of an automated microburst detection system which relies on data from a single Doppler radar. The comparison of radar and surface sensor data presented here investigates the possibility that some outflows may not be observable by radar due to: (1) low SNR (signal-to-noise ratio), (2) very shallow outflows for which the radar beam is scanning too high above the surface, (3) blockage of the beam, and/or (4) asymmetry in the surface outflow causing the radar to significantly underestimate the magnitude of the surface wind shear (Eilts and Doviak, 1987; GAO, 1987). Also addressed is the possibility that microbursts are not observed by the mesonet surface sensors because the spacing between stations is too great, or because the microburst outflow does not reach the surface due to a dense layer of cold air at the surface.

Thhis report investigates the observability of low-level wind shear events using Doppler weather radar through a comparison of radar and surface wind sensor data. The data was collected during 1986 in the Huntsville, AL area as part of the FAA Terminal Doppler Weather Radar (TDWR) development program. Radar data were collected by both an S-band radar (FL-2) and C-band radar (UND). Surface data were collected by a network of 77 weather sensors covering an area of enarly 1000 square km centered approximately 15 km to the northwest of the FL-2 radar site. The UND site was located at the approximate center of the surface sensor network. A list of 131 microbursts which impacted the surface sensor network is presented. Particular emphasis is on the 107 events for which both radar data and surface data where available. Of these events, 14 were not observed by the surface network, while two events were not identified as microbursts by radar. Possible explanations of these missed microburst identifications are presented. The first case was an instance of the radar viewing a weak, asymmetric event from an unfavorable viewing angle. The second case describes an extremely shallow microburst outflow occurring at a heigh too low to be observed by the lowest elevation scan of the radar. In each of these cases, the featured microburst was very weak and, although a microburst-strength differential velocity was not observable by radar, in both instances the divergent wind pattern associated with the event was clearly evident in the radar velocity data field. All microbursts which exhibited a differential velocoity of in excess of 13 m/s were identified by radar. No microbursts went unobserved as the result of insufficient signal return.