travel time measurements markus roth 1, laurent gizon 1, john g. beck 2 1 max-planck-institut für...
DESCRIPTION
Travel Time Measurements Cross-Correlation as function of distance and time lag Positive time-lag: outgoing-wave Negative time-lag: incoming wave Difference allows to conclude on flows, mean sound speed, etc... 1st ridge: one bounce 2nd ridge: two bounces 3rd ridge: three bounces... IntroductionMethodsNoiseResultsConclusions Observation at the equator, 90 day average Time-Distance DiagramTRANSCRIPT
Travel Time Measurements
Travel Time Measurements
Markus Roth1, Laurent Gizon1, John G. Beck2
1Max-Planck-Institut für Sonnensystemforschung2Stanford University
HELAS Workshop „Roadmap for Local Helioseismology“, September 25, 2006
Travel Time MeasurementsTime-Distance
Helioseismology
Introduction Methods Noise Results Conclusions
Δ
Source
Observer
Calculation of the Cross Correlation of the observed signals as function of travel-distance and time lag:
Observation of oscillation signal at two locations on the Sun
16.5°
Duvall et al (1993)
Travel Time Measurements
Cross-Correlation as function ofdistance and time lag
Positive time-lag: outgoing-wave
Negative time-lag: incoming wave
Difference allows to conclude onflows, mean sound speed, etc...
1st ridge: one bounce2nd ridge: two bounces3rd ridge: three bounces...
Introduction Methods Noise Results Conclusions
Observation at the equator, 90 day average
Time-Distance Diagram
Travel Time Measurements
Measuring Travel Times
Used Methods:
1) Fitting Garbor Wavelet (five parameters) ! phase travel time tph
2) Extra Cross-Correlation (Zhao & Jordan 1998, Gizon & Birch 2004 for stochastic sources), minimization of “badness of fit”:
yields travel-time measurement Reference cross-correlation: symmetrized long-term average
Correlation coefficient: 0.76
Introduction Methods Noise Results Conclusions
Travel Time Measurements
Measured Travel Times
Introduction Methods Noise Results Conclusions
Measuring the travel time difference between waves travelling
East – West
! sensitive to differential rotation
North – South
! sensitive to meridional flow
Travel Time Measurements
Wavelet Fits
Measuring travel times for 1st & 2nd bounce
Introduction Methods Noise Results Conclusions
1st bounce measurements are smooth
2nd bound measurements need more averagingclear due to signal/noise
Travel Time Measurements
Minimization Method
Introduction Methods Noise Results Conclusions
Advantages:
•Works on noisy data, too.
•Fast computation
•Stable results
Travel Time MeasurementsComparison of
PerformanceIntroduction Methods Noise Results Conclusions
Measured signal in the order of 1 sec(peak near 0 sec)
Wavelet fit:
• often no convergence
• outliers
• not Gaussian
Extra Cross-Correlation:
• No outliers
• Gaussian distribution
--- Garbor--- Extra CC
Travel Time Measurements
Variance Garbor Wavelets
Introduction Methods Noise Results Conclusions
Garbor Wavelets:Small errors at distances ¼ 10-20° & at latitutdes ¼ 0° Observational errors are largest for short and long distances, and for high latitudes
(consistent with P. Giles 1999, 90 day average)
North-South East-West
Travel Time Measurements
Variance Extra Cross-Correl.
Introduction Methods Noise Results Conclusions
North-South East-West
Extra Cross-Correlation:
As function of distance: dip at 10°, but more flat than with the Garbor waveletsAs function of latitude: minimal error at the equator, appears more flat
Travel Time Measurements
Noise Sources
Introduction Methods Noise Results Conclusions
Main Source: stochastic nature of solar oscillations (excitation, damping)
Additional sources (actually the main sources!):systematic noise sources (important for long averages, e.g. for precisions to measure meridional circulation, P-Angle, foreshortening, focus position of telescope)
Understanding of noise necessary for:
• correct understanding of measurements• solving the inverse problem, kernels (speed-up calculations)
Travel Time MeasurementsEstimating the
Covariance MatrixIntroduction Methods Noise Results Conclusions
Time-distance helioseismology:
Information on noise is given by the covariance matrix of the travel time measurements
x1
x2
x1´ x2´
Estimation: directly by averaging over many samples
´
Travel Time MeasurementsCovariance Matrix
Wavelet FitIntroduction Methods Noise Results Conclusions
Travel Time MeasurementsCovariance Matrix
Extra Cross-CorrelationIntroduction Methods Noise Results Conclusions
Travel Time MeasurementsCovariance Matrix
Wavelet FitIntroduction Methods Noise Results Conclusions
Travel Time MeasurementsCovariance Matrix
Extra Cross-CorrelationIntroduction Methods Noise Results Conclusions
Travel Time Measurements
Conclusions
Introduction Methods Noise Results Conclusions
The two methods are somehow different in their noise behaviour (1st & 2nd bounce)
! needs to be worked out in detail ! information gained about solar interior
Common: Covariance matrix is
• confined to a few degrees around the target distance
• confined to a few degrees around the target latitude (+ few oscillations)
Differences: between East-West and North-South measurments due to used points for averaging
! data dependence in covariance matrix