Publications
Birds mimicking microbursts on 2 June 1990 in Orlando, Florida
July 10, 1992
Project Report
Published in:
MIT Lincoln Laboratory Report ATC-184
Summary
During 1990 and 1991, the Terminal Doppler Weather Radar (TDWR) testbed collected Doppler radar measurements in Orlando, Florida in support of the TDWR Project. The main focus of the project is to develope algorithms that automatically detect wind shears such as microbursts anti gust fronts. While the primary goal of the TDWR is to detect scattering from raindrops, the sensitivity of the system allows for the detection of biological echoes as well. Previous research has shown that under certain conditions the scattering from birds and insects will lead to divergent signatures that mimic microbursts. This type, of pattern has been documented in Alabama (Rinehart, 1986), Illinois (Larkin and Quine, 1989), and Missouri (Evans, 1990). In the Alabama and Illinois events, a divergent pattern similar to a microburst was produced when a large number of birds departed in the early morning hours from an overnight roosting site. On 2 June 1990 in Orlando, Florida, there were 11 surface divergent signatures similar to microbursts detected by the TDWR testbed radar. The maximum differential velocity of these events ranged from 11 to 36 m/s, while the maximum reflectivity varied from 0 to 44 dBz. There was light rain in the area and low-reflectivity returns aloft; however, the reflectivity was more like low-reflectivity microbursts in Denver than high-reflectivity microbursts that generally are observed in Orlando. These divergences were not detected by the microburst algorithm since the TDWR site adaptation parameters have been adjusted to avoid issuing alarms for signatures such as those on 2 June. Detailed investigation was conducted of two events to verify that these were not actual microbursts. Single Doppler radar features identified in earlier observations of divergence signatures caused by birds in Alabama and Missouri, as well as features suggested by NEXRAD researchers, were considered. The results of the radar data analysis could not unequivocally determine that birds caused the divergent signatures. A microburst prediction model developed by Wolfson was applied to the data using sounding results from Cape Canaveral, Florida to determine whether the apparent velocities were consistent with current theories of microburst generation. This model analysis clearly indicated a nonweather-related cause for the divergent signatures observed on 2 June. We conclude from the microburst prediction analysis and certain oddities in the divergence radar signatures that birds probably accounted for these divergences.
Summary
During 1990 and 1991, the Terminal Doppler Weather Radar (TDWR) testbed collected Doppler radar measurements in Orlando, Florida in support of the TDWR Project. The main focus of the project is to develope algorithms that automatically detect wind shears such as microbursts anti gust fronts. While the primary goal of...
READ MORE
Airport Surveillance Radar (ASR-9) Wind Shear Processor - 1991 Test at Orlando, Florida
June 1, 1992
Project Report
Published in:
MIT Lincoln Laboratory Report ATC-189
Summary
An operational test of a Wind Shear Processor (WSP) add-on to the Federal Aviation Administration's airport surveillance radar (ASR-9) took place at Orlando International Airport during July and August 1991. The test allowed for both quantitative assessment of the WSP's signal processing and wind shear detection algorithms and for feedback from air traffic controllers and their supervisors on the strengths and weaknesses of the system. Thunderstorm activity during the test period was intense; low-altitude wind shear impacted the runways or approach/departure corridors on 40 of the 53 test days. As in previous evaluations of the WSP in the southeastern United States, microburst detection performance was very reliable. Over 95% of the strong microbursts that affected the Orlando airport during the test period were detected by the system. Gust front detection during the test, while operationally useful, was not as reliable as it should have been, given the quality of gust front signatures in the base reflectivity and radial velocity data from the WSP. Subsequent development of a Machine Intelligent gust front algorithm has resulted in significantly improved detection capability. Results from the operational test are being utilized in ongoing refinement of the WSP.
Summary
An operational test of a Wind Shear Processor (WSP) add-on to the Federal Aviation Administration's airport surveillance radar (ASR-9) took place at Orlando International Airport during July and August 1991. The test allowed for both quantitative assessment of the WSP's signal processing and wind shear detection algorithms and for feedback...
READ MORE
A comparison of the performance of two gust front detection algorithms using a length-based scoring technique
May 8, 1992
Project Report
Published in:
MIT Lincoln Laboratory Report ATC-185
Summary
The Terminal Doppler Weather Radar (TDWR) Gust Front Algorithm provides, as products, estimates of the current locations of gust fronts, their future locations, the wind speed and sirection behind the gust fronts, and the wind shear hazard to landing or departing aircraft. These products are used by air traffic controllers and supervisors to warn pilots of potentially hazardous wind shears during take-off and landing and to plan runway reconfigurations. Until recently, an event-based scoring system was used to evaluate the performance of the algorithm. With the event-based scoring scheme, if any part of a gust front length was detected, a valid detection was declared. Unfortunately, this scheme gave no indication of how much of the gust front length was detected; nor could the probabilities be easily related to the probability of issuing a wind shear alert for a specific approach or departure path which was being impacted by a gust front. To make the scoring metric more nearly reflect the operational use of the product, a new length-based scoring scheme was devised. This scheme computes the length of the gust front detected by the algorithm. When computed over a large number of gust fronts, this length-based scoring scheme yields the probability that any part of the gust front will be detected. As improvements to the algorithm increase the length detected, the probability of detecting any part of a gust front increases. In particular, an improved algorithm means an increased probability of correctly issuing wind shear alerts for the runways impacted by a gust front, and length-based scoring is a more accurate technique for assessing this probability of detection. This paper describes the length-based scoring scheme and compares it with event-based scoring of the algorithm's gust front detection and forecast performance. The comparison of the scoring methods shows that recent enhancements to the gust front algorithm provide a substantial, positive impact on performance.
Summary
The Terminal Doppler Weather Radar (TDWR) Gust Front Algorithm provides, as products, estimates of the current locations of gust fronts, their future locations, the wind speed and sirection behind the gust fronts, and the wind shear hazard to landing or departing aircraft. These products are used by air traffic controllers...
READ MORE
Two-talker pitch tracking for co-channel talker interference suppression
April 30, 1992
Technical Report
Published in:
MIT Lincoln Laboratory Report TR-951
Topic:
R&D area:
Summary
Almost all co-channel talker interference suppression systems use the difference in the pitches of the target and jammer speakers to suppress the jammer and enhance the target. While joint pitch estimators outputting two pitch estimates as a function of time have been proposed, the task of proper assignment of pitch to speaker (two-talker pitch tracking) has proven difficult. This report describes several approaches to the two-talker pitch tracking problem including algorithms for pitch track interpolation, spectral envelope tracking, and spectral envelope classification. When evaluated on an all-voiced two-talker database, the best of these new tracking systems correctly assigned pitch 87% of the time given perfect joint pitch estimation.
Summary
Almost all co-channel talker interference suppression systems use the difference in the pitches of the target and jammer speakers to suppress the jammer and enhance the target. While joint pitch estimators outputting two pitch estimates as a function of time have been proposed, the task of proper assignment of pitch...
READ MORE
Summary of triple Doppler data, Orlando 1991
April 7, 1992
Project Report
Published in:
MIT Lincoln Laboratory Report ATC-186
Summary
Under Federal Aviation Administration (FAA) sponsorship, Lincoln Laboratory conducted an aviation weather hazard measurement and operational demonstration program during the summer of 1991 near the Orlando International Airport. Three Doppler radars were sited in a triangle around the airport, allowing triple Doppler coverage of thunderstorms and microbursts occurring there. This report contains a summary of all of the microburst producing thunderstorms that occurred within the triple Doppler region that were scanned in a coordinated fashion, during the months of June, July, August, and September, 1991. Statistics on the microburst events are presented to give an overall picture of the available data for use in analysis. The bulk of the report consists of detailed information about each triple Doppler day, including the time, location, and strength of microbursts within the triple Doppler period as well as the availability of data from supporting sensors including the ASR-9-WSR Doppler radar, radiosondes, LLWAS, Mesonet, AWOS, instrumented aircraft, ACARS, interferometer, and corona points.
Summary
Under Federal Aviation Administration (FAA) sponsorship, Lincoln Laboratory conducted an aviation weather hazard measurement and operational demonstration program during the summer of 1991 near the Orlando International Airport. Three Doppler radars were sited in a triangle around the airport, allowing triple Doppler coverage of thunderstorms and microbursts occurring there. This...
READ MORE
Doppler mean velocity estimation - small sample analysis and a new estimator
March 25, 1992
Technical Report
Published in:
MIT Lincoln Laboratory Report TR-942
Summary
Optimal Doppler velocity estimation, under the constraint of small sample size, is explored for a standard Gaussian signal measurement model and thematic maximum likelihood (ML) and Bayes estimation. Because the model considered depends on a vector parameter [velocity, spectrum width, and signal-to-noise ratio (SNR)], the exact formulation of an ML or Bayes solution involves a system of equations that is neither uncoupled nor explicit in form. Historically, iterative methods have been the most suggested approach to solving the required equations. In addition to being computationally intensive, it is unclear whether iterative methods can be constructed to perform well given a small-sample size and low signal strength. This report takes a different approach and seeks to construct approximate (ML and Bayes) estimators based on the notion of using constrained adaptive models to deal with nuisance parameter removal. A Monte Carlo simulation is used to determine small-sample estimator statistics and to demonstrate true performance bounds in the case of known nuisance values. Performance comparisons between these optional forms and other standard estimators [pulse pairs (PP) and a frequency domain (WP) method] are presented. Performance sensitivity of the optimal algorithms, with respect to uncertainity in the values of model nuisance parameters, is explored.
Summary
Optimal Doppler velocity estimation, under the constraint of small sample size, is explored for a standard Gaussian signal measurement model and thematic maximum likelihood (ML) and Bayes estimation. Because the model considered depends on a vector parameter [velocity, spectrum width, and signal-to-noise ratio (SNR)], the exact formulation of an ML...
READ MORE
An integrated speech-background model for robust speaker identification
March 23, 1992
Conference Paper
Published in:
Proc. IEEE Int. Conf. on Acoustics, Speech and Signal Processing, ICASSP, Vol. 2, 23-26 March 1992, pp. 185-188.
Topic:
R&D area:
Summary
This paper examines a procedure for text independent speaker identification in noisy environments where the interfering background signals cannot be characterized using traditional broadband or impulsive noise models. In the procedure, both the speaker and the background processes are modeled using mixtures of Gaussians. Speaker and background models are integrated into a unified statistical framework allowing the decoupling of the underlying speech process from the noise corrupted observations via the expectation-maximization algorithm. Using this formalism, speaker model parameters are estimated in the presence of the background process, and a scoring procedure is implemented for computing the speaker likelihood in the noise corrupted environment. Performance is evaluated using a 16 speaker conversational speech database with both "speech babble" and white noise background processes.
Summary
This paper examines a procedure for text independent speaker identification in noisy environments where the interfering background signals cannot be characterized using traditional broadband or impulsive noise models. In the procedure, both the speaker and the background processes are modeled using mixtures of Gaussians. Speaker and background models are integrated...
READ MORE
A speech recognizer using radial basis function neural networks in an HMM framework
March 23, 1992
Conference Paper
Published in:
ICASSP'92, Proc. IEEE Int. Conf. on Acoustics, Speech and Signal Processing, Vol. 1, Speech Processing 1, 23-26 March 1992, pp. 629-632.
Topic:
R&D area:
Summary
A high performance speaker-independent isolated-word speech recognizer was developed which combines hidden Markov models (HMMs) and radial basis function (RBF) neural networks. RBF networks in this recognizer use discriminant training techniques to estimate Bayesian probabilities for each speech frame while HMM decoders estimate overall word likelihood scores for network outputs. RBF training is performed after the HMM recognizer has automatically segmented training tokens using forced Viterbi alignment. In recognition experiments using a speaker-independent E-set database, the hybrid recognizer had an error rate of 11.5% compared to 15.7% for the robust unimodal Gaussian HMM recognizer upon which the hybrid system was based. The error rate was also lower than that of a tied-mixture HMM recognizer with the same number of centers. These results demonstrate that RBF networks can be successfully incorporated in hybrid recognizers and suggest that they may be capable of good performance with fewer parameters than required by Gaussian mixture classifiers.
Summary
A high performance speaker-independent isolated-word speech recognizer was developed which combines hidden Markov models (HMMs) and radial basis function (RBF) neural networks. RBF networks in this recognizer use discriminant training techniques to estimate Bayesian probabilities for each speech frame while HMM decoders estimate overall word likelihood scores for network outputs...
READ MORE
Initialization for improved IIR filter performance
March 1, 1992
Journal Article
Published in:
IEEE Trans. Signal Process., Vol. 40, No. 3, March 1992, pp. 543-550.
Summary
A new method for initializing the memory registers of IIR filters is introduced. In addition to providing improved performance as compared to other methods of initialization, this method is unique in that it makes no a priori assumptions regarding the input-signal content. Therefore, this method applies equally well to a variety of IIR filter designs and applications. The method is best suited for signal-processing applications in which "batch" processing of the data is used. However, sequential processing can be accommodated when delays at the beginning of a processing segment can be tolerated.
Summary
A new method for initializing the memory registers of IIR filters is introduced. In addition to providing improved performance as compared to other methods of initialization, this method is unique in that it makes no a priori assumptions regarding the input-signal content. Therefore, this method applies equally well to a...
READ MORE
Shape invariant time-scale and pitch modification of speech
March 1, 1992
Journal Article
Published in:
IEEE Trans. Signal Process., Vol. 40, No. 3, March 1992, pp. 497-510.
Topic:
R&D area:
Summary
The simplified linear model of speech production predicts that when the rate of articulation is changed, the resulting waveform takes on the appearance of the original, except for a change in the time scale. The goal of this paper is to develop a time-scale modification system that preserves this shape-invariance property during voicing. This is done using a version of the sinusoidal analysis-synthesis system that models and independently modifies the phase contributions of the vocal tract and vocal cord excitation. An important property of the system is its capability of performing time-varying rates of change. Extensions of the method are applied to fixed and time-varying pitch modification of speech. The sine-wave analysis-synthesis system also allows for shape-invariant joint time-scale and pitch modification, and allows for the adjustment of the time scale and pitch according to speech characteristics such as the degree of voicing.
Summary
The simplified linear model of speech production predicts that when the rate of articulation is changed, the resulting waveform takes on the appearance of the original, except for a change in the time scale. The goal of this paper is to develop a time-scale modification system that preserves this shape-invariance...
READ MORE