What can we learn from climate data? [Elektronische Ressource] : methods for fluctuation, time, scale and phase analysis / von Douglas Maraun

Maraun

Sujets

• Mathematics

Informations

• Publié par: universitat_potsdam
• Publié le: 01 janvier 2006
• Nombre de lectures: 11
• Langue: English
• Poids de l'ouvrage: 3 Mo

Extrait

Institut fu¨r Physik· Arbeitsgruppe Nichtlineare Dynamik
What Can We Learn from Climate Data?
Methods for Fluctuation, Time/Scale and
Phase Analysis
Dissertation zur Erlangung des akademischen Grades
Doktor der Naturwissenschaften (Dr. rer. nat.)
in der Wissenschaftsdisziplin Nichtlineare Dynamik
Eingereicht an der
Mathematisch-Naturwissenschaftlichen Fakult¨at
der Universit¨at Potsdam
von
Douglas Maraun
Potsdam, im Juni 2006The following contributions have been published or submitted for publication as part of
this work or related to it:
D. Maraun, H.W. Rust, and J. Timmer. Tempting long-memory. On the interpretation
of DFA results. Nonlin. Proc. Geoph., 11(4):495–503, 2004.
D. Maraun and J. Kurths. Cross wavelet analysis. Significance testing and pitfalls. Nonlin.
Proc. Geoph., 11(4):505–514, 2004.
S.Rahmstorf,D.Archer,D.S.Ebel,O.Eugster,J.Jouzel, D.Maraun,U.Neu,G.A.Schmidt,
J.Severinghaus, A.J.Weaver, andJ.Zachos. Cosmicrays, carbondioxide, andclimate. Eos,
Trans. AGU, 85(4):38, 2004.
S.Rahmstorf,D.Archer,D.S.Ebel,O.Eugster,J.Jouzel, D.Maraun,U.Neu,G.A.Schmidt,
J. Severinghaus, A.J. Weaver, and J. Zachos. Reply to comment on ”cosmic rays, carbon
dioxide, and climate”. Eos, Trans. AGU, 85(48):511, 2004.
D. Maraun and J. Kurths. Epochs of phase coherence between El Nin˜o/ Southern Oscilla-
tionandIndianmonsoon. Geophys. Res. Lett.,32:L15709, doi:10.1029/2005GL023225, 2005.
J. Saynisch, J. Kurths, and D. Maraun. A Conceptual ENSO Modell under Realistic Noise
Forcing. accepted for publication in Nonlin. Proc. Geophys., 2006.
D. Maraun, J. Kurths, and M. Holschneider. Nonstationary Gaussian processes in wavelet
domain: Definitions, estimation and significance testing. submitted to Phys. Rev. E, 2006.
D. Maraun, B. Schaefli and M. Holschneider. Wavelet Spectral and Cross Spectral Analysis
in Hydrology. submitted for publication in the proceedings of the workshop “New Develop-
ments in Trend- and Extreme Value Analysis of Hydrometeorological Time Series”, Springer,
2006.
P.M. Crowley, D. Maraun and D. Mayes. How hard is the euro area core? A wavelet
analysis of growth cycles in Germany, France and Italy. submitted to Econ. J., 2006.To Antje, the friend of my years in Berliniv
Abbreviations often Used in this Thesis
ACF Auto correlation function
AIR All India Rainfall Index
CWT Continuous Wavelet Transformation
CDP Curvature defined phase
DFA Detrended Fluctuation Analysis
DWT Discrete Wavelet Transform
ENSO El Nin˜o/Southern Oscillation
HPA Hilbert phase analysis
ISM Indian summer monsoon
MODWT Maximum Overlap Discrete Wavelet Transformation
NAO North Atlantic Oscillation
SST Sea surface temperaturev
Notation
a Scale (in wavelet domain), a=1/f
A(t) Amplitude of an oscillation
A(.) Averaging operator
b Time (in wavelet domain)
C(s), C(r) Autocorrelation function
2COH (b,a) Squared wavelet coherency
CS(b,a) Wavelet cross spectrum
Cov Covariance
f Frequency
f(ω) Fourier multiplier
g(t) Wavelet
h(t) Reconstruction wavelet
H(.) Hilbert transformation
H Hurst exponent
H Positive halfplane
I Identity
m(a,b) Wavelet multiplier
M Inverse continuous wavelet transformation with reconstruction wavelet hh
P (λ) Probability distributionX
P Probability
ρ (x) Probability density function of the random variable XX
s(t) Time series
s Scale (in fluctuation analysis)
SE, SP Sensitivity, specificity
SNR Signal to noise ratio
S(b,a) Wavelet spectrum
Var Variance
W Continuous wavelet transformation with wavelet gg
x(t), x Time seriesi
γ Scaling exponent of the autocorrelation function
η noise
Φ(t) Phase of an oscillation
τ Characteristic time scaleD
ω Frequency, ω =2πf
ω Eigen-frequency of an oscillator, parameter of the Morlet wavelet0
¯. Complex conjugate
ˆ. Estimator, Fourier transformation (depending on context)
h.i Expectation valueviContents
Introduction xi
1 General Concepts 1
1.1 Random Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.1 Random Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.2 Multivariate Random Variables . . . . . . . . . . . . . . . . . . . . . . 2
1.1.3 Moments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.1.4 Quantiles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Stochastic Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2.1 Properties of Stochastic Processes . . . . . . . . . . . . . . . . . . . . 4
1.2.2 Gaussian Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 Estimators. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.3.1 Bias . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.3.2 Variance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.3.3 Consistency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.4 Significance Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.4.1 Statistical Errors and Power of the Test . . . . . . . . . . . . . . . . . 7
1.4.2 Sensitivity vs. Specificity . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.4.3 Multiple Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.5 Inverse Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2 Tempting Long Memory 9
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2 Autocorrelation Function of Stochastic Processes . . . . . . . . . . . . . . . . 10
2.2.1 Short Range Correlations . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.2.2 Long Range Correlations . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2.3 Discrimination between Short and Long Memory . . . . . . . . . . . . 11
2.3 Detrended Fluctuation Analysis (DFA) . . . . . . . . . . . . . . . . . . . . . . 11
2.3.1 The Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.3.2 Estimating the Strength of Long Range Correlations . . . . . . . . . . 12
2.4 Inference of Long Memory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.4.1 Two Example Processes . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.4.2 Establish Scaling of the Fluctuation Function . . . . . . . . . . . . . . 14
2.5 Inference of Short Memory . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.5.1 The Fluctuation Function of AR[1]-Processes . . . . . . . . . . . . . . 19
2.5.2 Finite Scaling of Short-Memory Processes . . . . . . . . . . . . . . . . 19
2.5.3 Shift of the finite scaling region . . . . . . . . . . . . . . . . . . . . . . 20
viiviii CONTENTS
2.6 Memory of Temperature Records . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.7 Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3 Continuous Wavelet Spectral Analysis 25
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.2 Mathematical Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.2.1 From Fourier to Wavelet Transformation . . . . . . . . . . . . . . . . . 26
3.2.2 Wavelets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.2.3 Continuous Wavelet Transformation . . . . . . . . . . . . . . . . . . . 28
3.2.4 Inverse Continuous Wavelet Transformation . . . . . . . . . . . . . . . 28
3.2.5 Reproducing Kernel . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.2.6 Choice of the Wavelet . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.2.7 Discrete Wavelet Transformation . . . . . . . . . . . . . . . . . . . . . 29
3.3 Motivation - The State of the Art . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.3.1 Conventional Definitions of Wavelet Spectral Measures . . . . . . . . . 30
3.3.2 Conventional Smoothing and Global Wavelet Spectrum . . . . . . . . 31
3.3.3 Conventional Pointwise Significance Testing . . . . . . . . . . . . . . . 31
3.3.4 Questions arising . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.4 Stochastic Processes defined in Wavelet Domain. . . . . . . . . . . . . . . . . 34
3.4.1 Stationary Gaussian Processes in Fourier Domain . . . . . . . . . . . . 34
3.4.2 Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.4.3 Spectral Measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.4.4 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.5 Estimating Wavelet Spectra . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.5.1 Spectral Estimators . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.5.2 Distribution, Variance and Bias of the Estimator . . . . . . . . . . . . 39
3.5.3 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.6 Significance Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.6.1 Pointwise Testing of the Wavelet Spectrum . . . . . . . . . . . . . . . 43
3.6.2 Areawise Testing of the Wavelet Spectrum. . . . . . . . . . . . . . . . 44
3.6.3 Sensitivity and Specificity of the Areawise Test . . . . . . . . . . . . . 46
3.6.4 Testing of Covarying Power . . . . . . . . . . . . . . . . . . . . . . . . 52
3