No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Bird identification on 1290MHz wind profiler data applying neural networks and neurofuzzy systems
Abstract
Migrating birds can severely affect data from wind profilers operating in the 1000 MHz range. Recent methods for removing bird contamination do not seem to solve the problem satisfactorily. Here, a new method, the Quantum Neurofuzzy Bird Identification and Removal Deck (NEURO-BIRD) is presented. The algorithm has an overall classification rate of over 90 % for birds, clear air returns, and rain echoes for single, one-second wind profiler spectra. Even with very heavy migration, high quality hourly winds can be obtained. Because the source of contamination of the spectra is unambiguously identified, bird data can be supplied for ornithological research. NEURO-BIRD is very fast and well suited for real-time applications.
... If such erroneous measurements are assimilated, then the quality of the forecasts is negatively impacted (Semple 2005; Cardinali 2009). This problem has been known for many years, see, for example, Ecklund et al. (1990), Barth et al. (1994), Wilczak et al. (1995, 1996), Engelbart et al. (1998), Richner and Kretzschmar (2001), Benjamin et al. (2004b), and Lehmann and Teschke (2008). Meanwhile, the increased performance of radar processors allows for the application of rather sophisticated clutter detection and filtering algorithms. ...
A new method is presented for Doppler spectral averaging that more reliably identifies the profiler radar return from clear air in the presence of contamination—for example, from migrating bird echoes. These very sensitive radars profile the wind in the atmosphere by detecting the very weak backscatter from index-of-refraction fluctuations caused by turbulence in clear air. But these radars also “see” the much stronger echoes from ground clutter (from trees and vehicles) and from fliers (aircraft, insects, and birds). Clutter and fliers have radar signatures in general that are different from atmospheric signatures. The new method uses this difference to extract atmospheric signals from contaminated Doppler spectra.
... Table 1 shows statistical results of the north and south cases. It is well known that radar wind profilers are contaminated by birds during their seasonal migration periods (Wilczak et al., 1995;Richner and Kretzschmar, 2001). Removing the bird signals from the registered signal has long been a challenge (Merrit, 1995;Jordan et al., 1997;Kretzschmar et al., 2003;Lehmann, 2012). ...
It is well known in the scientific community that some remote sensing
instruments assume that sample volumes present homogeneous conditions within
a defined meteorological profile. At complex topographic sites and under
extreme meteorological conditions, this assumption may be fallible depending
on the site, and it is more likely to fail in the lower layers of the
atmosphere. This piece of work tests the homogeneity of the wind field over
a boundary layer wind profiler radar located in complex terrain on the coast
under different meteorological conditions. The results reveal the
qualitative importance of being aware of deviations in this homogeneity
assumption and evaluate its effect on the final product. Patterns of
behavior in data have been identified in order to simplify the analysis of
the complex signal registered.
The quality information obtained from the homogeneity study under different
meteorological conditions provides useful indicators for the best
alternatives the system can offer to build wind profiles. Finally, the
results are also to be considered in order to integrate them in a quality
algorithm implemented at the product level.
... If such erroneous measurements are assimilated, then the quality of the forecasts is negatively impacted (Semple 2005;Cardinali 2009). This problem has been known for many years, see, for example, Ecklund et al. (1990), Barth et al. (1994), Wilczak et al. (1995Wilczak et al. ( , 1996, Engelbart et al. (1998), Richner and Kretzschmar (2001), Benjamin et al. (2004b), and Lehmann and Teschke (2008). ...
Intermittent clutter signals are frequently observed by radar wind profilers during the seasonal bird migration. A novel statistical filtering algorithm based on a simultaneous time-frequency analysis of the profiler's raw data was recently proposed to address shortcomings of existing methods. The foundation of this method is a Gabor frame expansion of the complex time series of the demodulated receiver voltage. In this paper, two objective criteria are suggested to obtain an optimal setup for the discrete Gabor frame expansion from the multitude of possibilities: first, the choice of almost-tight frames for a predefined maximum redundancy and second, the requirement that the analyzing bandwidth of the used Gaussian window function should provide a simultaneously sparse representation of both atmospheric signal components and intermittent clutter. The question of optimal sampling settings, especially dwell time, for a maximum reduction of bird interference is also discussed. Using data obtained during intense bird migration events it is shown that a combination of filtering and quality control of the result is required to prevent the occurrence of significant systematic and correlated errors in the final wind measurement.
... Table 1 shows statistical results of the north and south cases. It is well known that radar wind profilers are contaminated by birds during their seasonal migration periods (Wilczak et al., 1995;Richner and Kretzschmar, 2001). Removing the bird signals from the registered signal has long been a challenge (Merrit, 1995;Jordan et al., 1997;Kretzschmar et al., 2003;Lehmann, 2012). ...
It is well known amongst the scientific community that some remote
sensing instruments have assumed that sample volumes present homogeneous
conditions within a defined meteorological profile. At complex
topographic sites and under extreme meteorological conditions, this
assumption may be fallible depending on the site, and it is more likely
to fail in the lower layers of the atmosphere. This piece of work tests
the homogeneity of the wind field over a boundary layer wind profiler
radar located in a complex terrain on the coast, under different
meteorological conditions. The results reveal the qualitative importance
of being aware of the deviations of this homogeneity assumption and
evaluate its effect on the final product. Patterns of behaviour in data
have been identified in order to simplify the analysis of the complex
signal registered. The quality information obtained from
the homogeneity study under different meteorological conditions is
useful to look for the best alternatives the system can offer to build
wind profiles. Finally, the results are also to be considered in order
to integrate them in a quality algorithm implemented at product level.
... The problem of bird contamination has been addressed by many researchers (Merritt 1995, Jordan et al. 1997, Cornman et al. 1998, Richner and Kretzschmar 2001, to name a few); however, a practical solution to the problem is yet to be found. The radar wind profiler (RWP) works with extremely weak clear-air signals, and a reflection from even a single material object like a flying bird might be many times more powerful than the atmospheric signal. ...
The U. S. Department of Energy's Atmospheric Radiation Measurement Program is running a small network of 915-MHz radar wind profilers (RWPs) at its Southern Great Plains Cloud and Radiation Testbed site in northern Oklahoma and southern Kansas. Seasonal migration of passerines may cause significant interference with the operation of 915-MHz RWPs. The extent of this ''bird jamming'' depends on the radar's parameters, the place of deployment, the season, and the time of day. This poster presents a new diagnostic method for detecting possible bird contamination in RWP data, along with an evaluation of the method using a three-year data set for two RWPs.
This is a brief review of the main insights into bird migration provided
by radar. Radar is the main tool to study the flight behavior of
migratory birds under the influence of environmental factors, i.e., the
ecology of migratory flights, ranging from the large-scale pattern of
migration in relation to the distribution of land masses, geomorphology,
and weather systems down to the variation of flight behavior of single
birds in response to leading lines, obstacles, particular atmospheric
conditions, and flight phases.
Winds measured with 915- and 404-MHz wind profilers are frequently found to have nonrandom errors as large as 15 m s⁻¹ when compared to simultaneously measured rawinsonde winds. Detailed studies of these errors which occur only at night below about 4 km in altitude and have a pronounced seasonal pattern, indicate that they are due to the wind profilers' detection of migrating songbirds (passerines). Characteristics of contaminated data at various stages of data processing are described, including raw time series, individual spectra, averaged spectra, 30- or 60-s moments, 3- or 6-min winds, and hourly averaged winds. An automated technique for the rejection of contaminated data in historical datasets, based on thresholding high values of rnoment-level reflectivity and spectral width, is shown to be effective. Techniques designed for future wind profiter data acquisition systems are described that show promise for rejecting bird echoes, with the additional capability of being able to retrieve the true wind velocity in many instances. Finally, characteristics of bird migration revealed by wind profilers are described, including statistics of the spring (March–May) 1993 migration season determined from the 404-MHz Wind Profiler Demonstration Network (WPDN). During that time, contamination of moment data occurred on 43% of the nights monitored.
This paper introduces quantum neural networks (QNNs), a class of
feedforward neural networks (FFNNs) inherently capable of estimating the
structure of a feature space in the form of fuzzy sets. The hidden units
of these networks develop quantized representations of the sample
information provided by the training data set in various graded levels
of certainty. Unlike other approaches attempting to merge fuzzy logic
and neural networks, QNNs can be used in pattern classification problems
without any restricting assumptions such as the availability of a priori
knowledge or desired membership profile, convexity of classes, a limited
number of classes, etc. Experimental results presented here show that
QNNs are capable of recognizing structures in data, a property that
conventional FFNNs with sigmoidal hidden units lack
A new method is presented for Doppler spectral averaging that more reliably identifies the profiler radar return from clear air in the presence of contamination—for example, from migrating bird echoes. These very sensitive radars profile the wind in the atmosphere by detecting the very weak backscatter from index-of-refraction fluctuations caused by turbulence in clear air. But these radars also “see” the much stronger echoes from ground clutter (from trees and vehicles) and from fliers (aircraft, insects, and birds). Clutter and fliers have radar signatures in general that are different from atmospheric signatures. The new method uses this difference to extract atmospheric signals from contaminated Doppler spectra.
New algorithms for removing ground and intermittent clutter contamination from wind profiler data are presented. The techniques use wavelet transforms to "filter" the contribution of clutter to the wind profiler signals. Examples of typical clutter contamination using simulated and actual signals are presented. The corresponding Doppler spectra before and after wavelet filtering are compared. The authors find that wavelet filtering can reduce the clutter-to-clear-air signal power by as much as 40 dB, even when the clutter and clear-air peaks cannot be resolved in the Doppler spectrum. The enhancement in clear-air signal detectability in the presence of clutter is due to the more efficient separation of clutter and clear-air energy in the wavelet domain.
An abstract is not available.
This paper rigorously establishes that standard multilayer feedforward networks with as few as one hidden layer using arbitrary squashing functions are capable of approximating any Borel measurable function from one finite dimensional space to another to any desired degree of accuracy, provided sufficiently many hidden units are available. In this sense, multilayer feedforward networks are a class of universal approximators.
The Effect of Migrating Birds on 1290-MHz Wind Profiler Data
- Kretzschmar
Kretzschmar, R.: The Effect of Migrating Birds on 1290-MHz Wind Projler Data. Semester Thesis, Institute for Atmospheric Science (LAPETH), ETH, CH-8093 Ztirich, 3lp., 1997.
A statistical averaging method for wind profiler Doppler spectra Fuzzy Neural Processing for Pattern Recogniiion Karayiannis: Quantum Neural Networks (QNN's): Inherently Fuzzy Feedforward Neural Networks
- R Kretzschmar
- I290-Mhz Neuro-Bra
- Profiler Merritt
- D A Purushothaman
- G Purushothaman
- N B M Ralph
- B L Weber
- D A Merritt
Kretzschmar, R.: Quantum Neurofiazy Bird Removal Algorithm (NEURO-BRA) for I290-MHz Wind Profiler Data. Diploma Thesis, Institute for Atmospheric Science (LAPETH), ETH, CH-8093 Ztirich, 77p., 1998. Masters, T.: Practical Neural Neiwork Recipes in C+ +, Academic Press, San Diego, 493p., 1993. Merritt, D.A.: A statistical averaging method for wind profiler Doppler spectra. J. Atmos. Ocean. TechnoL, 12,985-995, 1995. Purushothaman, G.: Fuzzy Neural Processing for Pattern Recogniiion. Master thesis, Dept. Elech-. Computer Eng., Univ. Houston, 1994. Purushothaman, G. and N.B. Karayiannis: Quantum Neural Networks (QNN's): Inherently Fuzzy Feedforward Neural Networks. IEEE Trans. Neural Networks, 8679-693, 1997. SNNS Group, Institute for Parallel and Distributed High-Performance Systems (IPVR), University of Stuttgart, Breitwiesenstrasse 20-22, D-70565 Stuttgart, Germany. The SNNS simulator can be downloaded from: Rp://fip.informatik.uni-stuttgart.de/pub/SNNS/ Wilczak, J.M., R.G. Strauch, F.M. Ralph, B.L. Weber, D.A. Merritt, J.R.
The study of bird migration by radar, Part 2: Major achievements
- B Bruderer
- K Hornik
- M Stinchcombe
- H White
Bruderer, B., The study of bird migration by radar, Part 2: Major achievements. Naturwissenschaflen, 84,45-54, 1997. Hornik, K., M. Stinchcombe, and H. White: Multilayer Feedforward Networks are universal Approximators. Neural Networks, 2, 359-366, 1989. reprinted in: White, H. (ed.): ArtiJcial neural networks. Approximation and learning theory, Blackwell Publishers, Oxford, 329p., 1992.
Quantum Neurofuzzy Bird Removal Algorithm (NEUROBRA) for 1290-MHz Wind Profiler Data
- Kretzschmar