Publications

In this paper we introduce a new baseline for language-independent text difficulty assessment applied to the Interagency Language Roundtable (ILR) proficiency scale. We demonstrate that reading level assessment is a discriminative problem that is best-suited for regression. Our baseline uses z-normalized shallow length features and TF-LOG weighted vectors on bag-of-words for Arabic, Dari, English, and Pashto. We compare Support Vector Machines and the Margin-Infused Relaxed Algorithm measured by mean squared error. We provide an analysis of which features are most predictive of a given level.

The Integrated Terminal Weather System (ITWS), currently in development by the FAA, will produce a fully-automated integrated terminal weather information system to improve the safety, efficiency and capacity of terminal area aviation operations. The ITWS will acquire data from FAA and National Weather Service sensors as well as from aircraft in flight in the terminal area. The ITWS will provide products to Air Traffic personnel that are immediately usable without further meteorological interpretation. These products include current terminal area weather and short-term (0-30 minute) predictions of significant weather phenomena. The Terminal Doppler Weather Radar (TDWR) will serve as a principle sensor providing data to a number of the ITWS algorithms. One component of the ITWS will be an algorithm for detecting gust fronts and wind shifts. A gust front is the leading edge of a cold air outflow from a thunderstorm. The outflow, which is deflected at the ground, may propagate many miles ahead of the generating thunderstorm, and may persist as an outflow boundary long after the original storm has dissipated. Gust fronts can have a significant impact on air terminal operations since they often produce pronounced changes in wind speed and direction, forcing a change in active runway configuration and rerouting of aircraft within in the terminal airspace. In addition, wind shear, turbulence, and cross-winds along the frontal boundary pose significant safety hazards to departing and landing aircraft. Reliable detection and forecasting of gust fronts and wind shifts will both improve air safety and reduce costly delays. Lincoln Laboratory has developed an Initial Operational Capability (IOC) Machine Intelligent Gust Front Algorithm (MIGFA) for the ITWS which currently utilizes TDWR and LL WAS or ASOS anemometer data and makes use of new techniques of knowledge-based signal processing originally developed in the context of automatic target recognition [Verly, 1989]. Extensions to the IOC to incorporate additional sensor or product data available under the ITWS (e.g., NEXRAD, terminal winds) are currently under development. MIGFA was first developed for the Airport Surveillance Radar with Wind Shear Processor (ASR-9 WSP). Its design and performance have been documented in previous reports by the authors [Delanoy 1993a]. This paper focuses on the design of the more recently developed TDWR MIGFA and its extension and adaptation to the ITWS (a more detailed description of the TDWR MIGFA can be found in Troxel [1994]). An overview of the signal processing techniques used for detection and tracking is presented, as well as a brief discussion of the wind analysis methods used to arrive at the wind shift and wind shear estimates. Quantitative performance analyses using data collected during recent field testing in Orlando, FL and Memphis, TN are presented. Test results show that MIGFA substantially outperforms the gust front detection algorithm used in current TDWR systems [Hermes, 1993] (MIGFA is currently under consideration as an upgrade option for TDWR).

Automated gust front detection is an important component of the Airport Surveillance Radar with Wind Shear Processor (ASR-9 WSP) and Terminal Doppler Weather Radar (TDWR) systems being developed for airport terminal areas. Gust fronts produce signatures in Doppler radar imagery which are often weak, ambiguous, or conditional, making detection and continuous tracking of gust fronts challenging. Previous algorithms designed for these systems have provided only modest performance when compared against human observations. A Machine Intelligent Gust Front Algorithm (MIGFA) has been developed that makes use of two new techniques of knowledge-based signal processing originally developed in the context of automatic target recognition. The first of these, functional template correlation (FTC), is a generalized matched filter incorporating aspects of fuzzy set theory. The second technique is the use of "interest" as a medium for pixel-level data fusion. MIGFA was first developed for the ASR-9 WSP system. Its design and performance have been documented in a number of earlier reports. This paper focuses on the more recently developed TDWR MIGFA, describing the signal-processing techniques used and general algorithm design. A quantitative performance analysis using data collected during recent real-time testing of the TDWR MIGFA in Orlando, Florida is also presented. Results show that MIGFA substantially outperforms the gust front detection algorithm used in current TDWR systems.

For reasons of aviation safety and airport operations efficiency, gust front detection and tracking is an important product of Doppler weather radars developed for use in airport terminal areas. Previous gust front algorithms, which have relied on the detection of one or two conspicuous signatures in Doppler radar imagery, have worked reasonably well in images generated by the high-resolution, pencil-beam Terminal Doppler Weather Radar (TDWR). The latest Airport Surveillance Radar, enhanced with a Wind Shear Processor (ASR-9 WSP), is being developed as a less expensive alternative weather radar. Although gust fronts are visible to human observers in ASR-9 WSP imagery, the lower sensitivity and less reliable Doppler measurements of this radar make automated gust front detection a much more challenging problem. Using machine intelligence and knowledge-based signal processing techniques developed in the context of automatic target recognition, a Machine Intelligent Gust Front Algorithm (MIGFA) has been constructed that is radically different from the previous algorithms. Developed initially for use with ASR-9 WSP data, MIGFA substantially outperforms a state-of-the-art gust front detection algorithm based on earlier approaches. These results also indirectly suggest that MIGFA performance may be nearly as good as human performance. Preliminary results of an operational test period (two months, approximately 15000 scans processed) are presented.

We have developed a computer program that automates rudimentary air traffic control (ATC) planning and decision-making functions. The ability to plan, make decisions, and act on them makes this experimental program qualitatively different from the more clerical ATC software currently in use. Encouraging results were obtained from tests involving simple scenarios used to train air traffic controllers.

This paper describes an artificial intelligence-based approach for automated recognition of wind shear hazards. The design of a prototype system for recognizing low-altitude wind shear events from Doppler radar displays is presented. This system, called WXI, consists of a conventional expert system augmented by a specialized capability for processing radar images. The radar image processing component of the system employs numerical and computer vision techniques to extract features from radar data. The expert system carries out symbolic reasoning on these features using a set of heuristic rules expressing meteorological knowledge about wind shear recognition. Results are provided demonstrating the ability of the system to recognize microburst and gust front wind shear events.