Effect of data gaps: Comparison of different spectral analysis methods Effect of data gaps: Comparison of different spectral analysis methods Catalin Negrea.

Slides:



Advertisements
Similar presentations
Simulation and Evaluation of Various Block Assignments Evaluation of multiple carriers deployed in a channel block evaluation criteria section.
Advertisements

Yinyin Yuan and Chang-Tsun Li Computer Science Department
Gravitational Wave Astronomy Dr. Giles Hammond Institute for Gravitational Research SUPA, University of Glasgow Universität Jena, August 2010.
Pattern Recognition and Machine Learning
FilteringComputational Geophysics and Data Analysis 1 Filtering Geophysical Data: Be careful!  Filtering: basic concepts  Seismogram examples, high-low-bandpass.
Engineering Mathematics Class #15 Fourier Series, Integrals, and Transforms (Part 3) Sheng-Fang Huang.
1 Chapter 16 Fourier Analysis with MATLAB Fourier analysis is the process of representing a function in terms of sinusoidal components. It is widely employed.
Page 0 of 34 MBE Vocoder. Page 1 of 34 Outline Introduction to vocoders MBE vocoder –MBE Parameters –Parameter estimation –Analysis and synthesis algorithm.
Digital Signal Processing
Constellations Demystified Presented by Sunrise Telecom Broadband … a step ahead.
CHAPTER 4 Noise in Frequency Modulation Systems
“Physics at the End of the Galactic Cosmic-Ray Spectrum” Aspen, CO 4/28/05 Diffusive Shock Acceleration of High-Energy Cosmic Rays The origin of the very-highest-energy.
Properties of continuous Fourier Transforms
Discrete Fourier Transform(2) Prof. Siripong Potisuk.
Chapter 5 Solutions for Interacting Waves Using A MCM 5.1 Governing Equations and Hierarchy Eq.s 5.2 An Example of Applying A Mode Coupling Method (MCM)
Spectral Analysis Spectral analysis is concerned with the determination of the energy or power spectrum of a continuous-time signal It is assumed that.
Wave Coupling Between the Lower Atmosphere and Thermosphere: Solar Cycle Influences 1 Jeffrey M. Forbes 1, Sean L. Bruinsma 2, Maura E. Hagan 3, and Xiaoli.
Zbigniew LEONOWICZ, Tadeusz LOBOS Wroclaw University of Technology Wroclaw University of Technology, Poland International Conference.
On the Accuracy of Modal Parameters Identified from Exponentially Windowed, Noise Contaminated Impulse Responses for a System with a Large Range of Decay.
Lecture 1 Signals in the Time and Frequency Domains
Fourier (1) Hany Ferdinando Dept. of Electrical Eng. Petra Christian University.
DOLPHIN INTEGRATION TAMES-2 workshop 23/05/2004 Corsica1 Behavioural Error Injection, Spectral Analysis and Error Detection for a 4 th order Single-loop.
Sensitivity System sensitivity is defined as the available input signal level Si for a given (SNR)O Si is called the minimum detectable signal An expression.
Understanding ADC Specifications September Definition of Terms 000 Analogue Input Voltage Digital Output Code FS1/2.
Intermittency beyond the ecliptic plane Anna Wawrzaszek, Marius Echim, Wiesław M. Macek, Roberto Bruno Mamaia, 6-13 September 2015 (1) Space Research Centre.
Noise characteristics Reference: [4] The signal-to-noise ratio is the measure for the extent to which a signal can be distinguished from the background.
Blue: Histogram of normalised deviation from “true” value; Red: Gaussian fit to histogram Presented at ESA Hyperspectral Workshop 2010, March 16-19, Frascati,
Acoustic Analysis of Speech Robert A. Prosek, Ph.D. CSD 301 Robert A. Prosek, Ph.D. CSD 301.
Speech Enhancement Using Spectral Subtraction
Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display. 1 Chapter 19.
Module 2 SPECTRAL ANALYSIS OF COMMUNICATION SIGNAL.
Physics 114: Exam 2 Review Lectures 11-16
SPATIAL AND TEMPORAL MONITORING OF THE INTERMITTENT DYNAMICS IN THE TERRESTRIAL FORESHOCK Péter Kovács, Gergely Vadász, András Koppán 1.Geological and.
Chapter 8 Frequency-Response Analysis
Seismic Reflection Data Processing and Interpretation A Workshop in Cairo 28 Oct. – 9 Nov Cairo University, Egypt Dr. Sherif Mohamed Hanafy Lecturer.
1 Searching for Periodic Gene Expression Patterns Using Lomb-Scargle Periodograms Critical Assessment.
Eeng Chapter 5 AM, FM, and Digital Modulated Systems  Phase Modulation (PM)  Frequency Modulation (FM)  Generation of PM and FM  Spectrum of.
Feodor Vybornov, Alexander Pershin, Alexander Rakhlin, Olga Sheiner Features of modern diagnostic methods of ionospheric turbulence Radiophysical Research.
NEW POWER QUALITY INDICES Zbigniew LEONOWICZ Department of Electrical Engineering Wroclaw University of Technology, Poland The Seventh IASTED International.
Part 4 Chapter 16 Fourier Analysis PowerPoints organized by Prof. Steve Chapra, University All images copyright © The McGraw-Hill Companies, Inc. Permission.
Basic Time Series Analyzing variable star data for the amateur astronomer.
Lecture#10 Spectrum Estimation
Z bigniew Leonowicz, Wroclaw University of Technology Z bigniew Leonowicz, Wroclaw University of Technology, Poland XXIX  IC-SPETO.
Basic Image Processing February 1. Homework Second homework set will be online Friday (2/2). First programming assignment will be online Monday (2/5).
Intermittency Analysis and Spatial Dependence of Magnetic Field Disturbances in the Fast Solar Wind Sunny W. Y. Tam 1 and Ya-Hui Yang 2 1 Institute of.
Chapter 4. Angle Modulation. Overview Angel modulation –The angle of the carrier wave is varied according to the information-bearing signal Lesson 1 :
Signal Analyzers. Introduction In the first 14 chapters we discussed measurement techniques in the time domain, that is, measurement of parameters that.
The Spectrum n Jean Baptiste Fourier ( ) discovered a fundamental tenet of wave theory.
BE/ABP/LIS section meeting - N. Mounet, B. Salvant and E. Métral - CERN/BE-ABP-LIS - 07/09/20091 Fourier transforms of wall impedances N. Mounet, B. Salvant.
The Frequency Domain Digital Image Processing – Chapter 8.
Analyzing circadian expression data by harmonic regression based on autoregressive spectral estimation Rendong Yang and Zhen Su Division of Bioinformatics,
Eeng Chapter 2 Discrete Fourier Transform (DFT) Topics:  Discrete Fourier Transform. Using the DFT to Compute the Continuous Fourier Transform.
Frequency Domain Representation of Biomedical Signals.
Eeng Chapter 5 AM, FM, and Digital Modulated Systems  Phase Modulation (PM)  Frequency Modulation (FM)  Generation of PM and FM  Spectrum of.
Data statistics and transformation revision Michael J. Watts
Stats Methods at IC Lecture 3: Regression.
FFT corrections for tune measurements
Robert Anderson SAS JMP
National Mathematics Day
CS 591 S1 – Computational Audio
Spectral Analysis Spectral analysis is concerned with the determination of the energy or power spectrum of a continuous-time signal It is assumed that.
Part 4 Chapter 16 Fourier Analysis
M.KARTHIK (10F41D4307) Under the esteemed guidance of
ივანე ჯავახიშვილის სახელობის
Fourier Analyses Time series Sampling interval Total period
Filtering Geophysical Data: Be careful!
Introduction to Frequency Domain TIPL 4301 TI Precision Labs – ADCs
Fourier Analyses Time series Sampling interval Total period
Lec.6:Discrete Fourier Transform and Signal Spectrum
A handbook on validation methodology. Metrics.
Presentation transcript:

Effect of data gaps: Comparison of different spectral analysis methods Effect of data gaps: Comparison of different spectral analysis methods Catalin Negrea 1,2,3 Costel Munteanu 1,4,5 1 Institute of Space Science, Magurele, Romania, 2 CIRES, University of Colorado at Boulder, 2 CIRES, University of Colorado at Boulder, 3 NOAA/SWPC, Boulder, CO, USA, 3 NOAA/SWPC, Boulder, CO, USA, 4 Department of Physics, University of Oulu, Finland 4 Department of Physics, University of Oulu, Finland 5 Department of Physics, University of Bucharest, Magurele, Romania 5 Department of Physics, University of Bucharest, Magurele, Romania September 2015 | Mamaia, ROMANIA Solar System Plasma Turbulence, Intermittency and Multifractals International Workshop and School Solar System Plasma Turbulence, Intermittency and Multifractals International Workshop and School

22 Basic Spectral Analysis September 2015 | Mamaia, ROMANIA Fast Fourier Transform

33 Basic Spectral Analysis September 2015 | Mamaia, ROMANIA

44 Methods Used September 2015 | Mamaia, ROMANIA

55 Tests on Synthetic Data Synthetic, continuous dataset: September 2015 | Mamaia, ROMANIA A superposition of sinusoids with added noise. To this “data” we apply a series of gaps and re-compute the Amplitude Spectrum using all four methods.

66 Tests on Synthetic Data In the case of a single large gap, ZTR and LST obtain good results, even for gaps of 90%. The drawback is a decrease of the Signal-to-Noise Ratio. FFT and DFT show a bias dependent on the gap size. This behavior of all four methods is the same for all four synthetic harmonics September 2015 | Mamaia, ROMANIA

77 Tests on Synthetic Data A more in-depth analysis shows a growth of the side-lobes associated with each harmonic as the gap size increases for the ZTR and LST. For applications where we need to find the amplitude of a dominant harmonic with a known frequency, LST and ZTR are likely to perform well even in the case of extremely large data gaps (~90%). This is achieved at the cost of an increase in side-lobe level September 2015 | Mamaia, ROMANIA

88 Another frequently encountered situation is the presence of multiple, small gaps of varying size and position September 2015 | Mamaia, ROMANIA Tests on Synthetic Data We simulate this scenario by applying gaps distributed in accordance with a gamma distribution. The resulting time series are analyzed with the four methods tested.

99 Tests on Synthetic Data Similarly, the ZTR and LST correctly report the amplitude of all four harmonics even for large gaps. In addition, the DFT results are similarly accurate. The tradeoff is lower SNR, but in this case due to random noise. The FFT again shows a bias dependent both on frequency and on the Total Gap Size September 2015 | Mamaia, ROMANIA Results for a Single Large Gap Results for a Multiple Small Gaps

1010 Tests on Venus Express Data A continuous magnetic field dataset taken by Venus Express in 2007 was also analyzed. The same methodology was used, applying different configurations of gaps to test the behavior of the 4 spectral analysis methods September 2015 | Mamaia, ROMANIA

1111 Tests on Venus Express Data When applying a single large gap to the data, the ZTR and LST results show a positive bias, while the FFT and DFT results show a negative bias September 2015 | Mamaia, ROMANIA

1212 Tests on Venus Express Data Integrating the Power Spectra gives a better idea of the behavior of each method as a function of gap size September 2015 | Mamaia, ROMANIA

1313 Tests on Venus Express Data In the case of multiple small gaps applied on Venus Express data, the results are surprising. The method that best preserves the shape (and slope) of the spectra is the FFT with linear interpolation. The DFT, ZTR and LST report biased spectra, with a frequency dependence to this bias September 2015 | Mamaia, ROMANIA

1414 Tests on Venus Express Data The behavior of the integral of the spectrum as a function of total gap percentage gives a better idea on the extent of this bias September 2015 | Mamaia, ROMANIA

1515 Conclusions There is no generally valid solution for the problems caused by data gaps. The method chosen will depend on the type of data analyzed and on the signal of interest. In the analysis of time series containing strong sinusoidal harmonics, the most accurate methods are found to be Lomb-Scargle and the Z- Transform. The reported amplitudes are accurate even in the presence of very large data gaps, with the price of a lower SNR. For more turbulent spectra, the appearance of side lobs and spectral noise makes the effect of data gaps more pronounced for the DFT, ZTR and LST methods than for FFT. Thus, our results indicate that, at least when the data includes gaps, FFT is the best method to approximate reliably the spectral slope of a signal recorded in a turbulent environment. Other, more complicated errors may also occur in the case of particular gap configurations September 2015 | Mamaia, ROMANIA

1616 Thank you! September 2015 | Mamaia, ROMANIA

1717 Confidence Test September 2015 | Mamaia, ROMANIA