A geomorhologic approach to estimate landslide hazard and risk in urban and rural areas - PowerPoint PPT Presentation

1 / 15
About This Presentation
Title:

A geomorhologic approach to estimate landslide hazard and risk in urban and rural areas

Description:

WP5: Temporal Scaling and Correlations. WP5: Temporal Scaling and Correlations ... stochastic differential equations, Mandelbrot cartoon Brownian motions, etc. ... – PowerPoint PPT presentation

Number of Views:67
Avg rating:3.0/5.0
Slides: 16
Provided by: faustog
Category:

less

Transcript and Presenter's Notes

Title: A geomorhologic approach to estimate landslide hazard and risk in urban and rural areas


1
(No Transcript)
2
Time Series
  • Many attributes, including
  • Frequency-size probability distribution of
    values. Ordering of values does not matter.
  • Pattern of the data series ordering
    matterscorrelations (persistence), long-range
    vs. short range

3
Long-range correlations
  • Many ways to measure the strength of (long-range)
    correlations of a time series, including
  • Fourier Power-Spectral Analysis
  • Detrended Fluctuation Analysis
  • Autocorrelation
  • Semivariogram Analysis
  • Hurst Rescaled-Range (R/S) Analysis
  • Wavelet Spectral Analysis
  • Interevent occurrence times

4
Fourier Power Spectral Analysis
S f b S power spectral density (Amplit
ude of Fourier coefficients)2 f frequency b
constant b 5 2D
b 0 White noise. Uncorrelated b 1 Pink
(1/f ) noise. Medium strength long-range
persistence. b 2 Brownian motion. Strong
long-range persistence, e.g. topography.
5
Power-Spectral Analysis
Data Series
6
Wavelet analysis of annual tree-ring widths for
bristle cone pines (pinus longaeva) on Campito
Mtn, California ITRDB USA 37N,118W Graybill
LaMarche See also papers by Harlan Ferguson.
7
Wavelet variance analysis of annual tree-ring
data for Campito Mtn, California (2111 BC to 1969
AD). Larger b indicates stronger long-range
persistence .
8
Measures of long-range persistence
  • Fourier Power-Spectral Analysis
  • bPS bPS52D
  • Detrended Fluctuation Analysis
  • aDFA bDFA 2aDFA1
  • Semivariogram Analysis
  • HaHausdorff Exponent bSemivariogram 2Ha1 0
    lt Ha lt 1
  • Hurst Rescaled-Range (R/S) Analysis
  • HuHurst Exponent bHurst 2Hu1 0 lt Hu lt 1
  • Wavelet Spectral Analysis bWavelet
  • Autocorrelation

9
Create Fractional Noises
  • Why? As benchmarks with given properties to
    examine the many techniques.
  • Fourier Filtering Fourier Filter a white noise
    so that is has a prescribed strength of
    long-range persistence
  • Others Discrete Wavelets, random additions,
    Surrogate data techniques, stochastic
    differential equations, Mandelbrot cartoon
    Brownian motions, etc.

10
Poster Performance Tests for Techniques that
Measure Long-Range Persistence
Poster Detrended Fluctuation Analysis
11
S f b (S power spectral density f
frequency b constant )
Fractional Gaussian Noises created by Fourier
Filtering
12
Fractional log-normal Noises, cv mean/s
1.0 created from the fGNs from Fourier Filtering
13
(No Transcript)
14
(No Transcript)
15
Discussion
  • Methods for synthetic time series construction
    (heavy tails?)
  • Techniques for quantifying (long-range)
    correlations in time series
  • Fitting data
  • Statistical tests (fitting data, hypothesis
    tests, error bars, etc.)
Write a Comment
User Comments (0)
About PowerShow.com