Tags:
create new tag
view all tags

Local Helioseismology Inversion Deployment

Science Testing Document

Input

  • The algorithm takes as its first input 3D FITS file, containing travel times for a given set of skip-distances. The fits file name is specified by the prefix -fi. Data of float type is accepted, there are no restrictions on size (subject to memory restrictions). The input FITS file remains unmodified.
  • The spatial resolution of the travel time data is specified using -ri keyword.
  • The algorithm takes as its second input 3D FITS file, containing corresponding sensitivity kernels. The fits file name is specified by the prefix -ki. Data of float type is accepted, there are no restrictions on size (subject to memory restrictions). The input FITS file remains unmodified.
  • The spatial resolution of the kernel data is specified using -rk keyword.

Optional Inputs

  • To specify the trade-off parameter for Multi-Channel Deconvolution use prefix -eps followed by the value
  • To specify the amount of horizontal regularisation for Multi-Channel Deconvolution use prefix -pp followed by the value. The value 0 (default) means no horizontal regularisation is applied.
  • For more options see documentation

Compilation and Execution:

code uses the following GNU Public License Libraries:

CFITSIO (for reading & writing FITS format files) http://heasarc.gsfc.nasa.gov/docs/software/fitsio/fitsio.html

FFTW (float version, i.e. libfftw3f.a library) http://www.fftw.org/, for Windows: http://www.fftw.org/install/windows.html

GSL (General Scientific Library) http://www.gnu.org/software/gsl/

  • gcc compilation: %gcc -o inv Inversion.c MCDInversion.c -lgsl -lgslcblas -lcfitsio /usr/local/lib/libfftw3f.a -I$(LIBS)/cfitsio/include
  • commandline execution:

* %./inv -fi input_data.fits -fo output -eps 2500. -ri 1.518 -ki kernels/JKernel.fits -kr 1.652
* %./inv -fi input_data.fits -fo output -eps 2500. -pp 1.e+1 -ri 1.51815 ki kernels/JKernel_1.518.fits -kr 1.518 -ss kernels/JKernel_ss.fits -di kernels/JKernel_zi.fits
* %./inv -fi input_data.fits -fo output -eps 2500. -pp 1.e+1 -ri 1.51815 -ki kernels/JKernel.fits -kr 1.652 -ak

AstroGrid workflow

  • AstroGrid workflow instructions:
    • Open AstroGrid workbench and click "Task Launcher"
    • Tasks: find application, specify files and variables as input, specify files as output, launch
    • Task Launcher search: helioseismology inversion or "Local Helioseismology Inversion"
    • Input:
    • Output:
      • OutputFile: iinv_SS_NOAA8038.fits (file reference; MySpace fits file)

Expected Output

  • final output:
    • output is 3D FITS file with the results of inversion, either sound speed perturbation or flow map depending on the choice of input data and sensitivity kernel. The depths are specified by the choice of skip-distances and kernels.
    • optional output - 4D FITS file with the resolution kernels for the inversion
    • optional output - 4D FITS file with noise covariance estimate

  • current level of completion:
    • the choice of regularisation is currently limited to slowness

  • limitations of algorithm:

Unit Testing

Classes with unit tests:

  • Inversion.c (contains currently 3 tests)
  • MCDInversion.c (contains currently 1 test)

unit test compilation
gcc -Wall allTests.c InvTest.c MCDInversion.c -lgsl -lgslcblas -lcfitsio /usr/local/lib/libfftw3f.a -I$(LIBS)/cfitsio/include -o runtest

Science Test Cases

See LocalHelioseismologyDeploymentResults

Case 1: Invert mean travel-times using Multi-Channel Deconvolution & Rytov approximation soundspeed sensitivity kernels

Description

Invert the mean travel-times using corresponding to the configuration sound-speed sensitivity kernels. The case is part of the general Science Case for obtaining the wavespeed Perturbation map.

Input

mean travel times & configuration file
corresponding sensitivity kernels (Rytov approximation)
Choice of horizontal smoothing parameter Choice of trade-off parameter

command line: ./inv -fi tt_Gabor_IO_NOAA8038_mean.fits -fo inv_SS_NOAA8038-ri 1.518 -kr 1.652 -krn kernels/JKernel.fits -eps 20000. -pp 70.

Output

Inversion of the travel-times (3D fits file) providing sound speed contrast as a function of position and depth.

Case 2: Invert travel-time differences using Multi-Channel Deconvolution & Born approximation flow sensitivity kernels

Description

Invert the travel-time differneces using corresponding to the configuration subsurface flow sensitivity kernels. The case is part of the general Science Case for obtaining the Subsurface flow map.

Input

3 travel time difference maps, correspondng to waves travelling in East-West, North South and In Out direction
configuration file describing geometry and filtering
corresponding sensitivity kernels (Born approximation) for each direction
Choice of horizontal smoothing parameter Choice of trade-off parameter
commandline execution:
./inv -fi tt_IO_NOAA8038_mean.fits -fo inv_IO_NOAA8038 -ri 1.518 -kr 1.518 -krn kernels/krnFlow_VZ.fits -abk -flow -eps 50. -pp 20.
./inv -fi tt_EW_NOAA8038_mean.fits -fo inv_EW_NOAA8038 -ri 1.518 -kr 1.518 -krn kernels/krnFlow_EW.fits -abk -flow -eps 50. -pp 20.
./inv -fi tt_NS_NOAA8038_mean.fits -fo inv_NS_NOAA8038 -ri 1.518 -kr 1.518 -krn kernels/krnFlow_NS.fits -abk -flow -eps 50. -pp 20.

Expected Output

Inversion of the travel-times (3D fits file) providing horizontal flow velocities as a function of position and depth.

Case 3: Noise Propagation in Inversion of Helioseismic Time-Distance Data

Description

Estimate the error in the inversion using data noise covariance matrix.

Input

mean travel times & configuration file
resampled slowness sensitivity kernels (Rytov approximation) JKernel_1.518.fits
data noise covariance ttnoiseCovar_128128_r1.518.fits

./inv -ki kernels/JKernel_1.518.fits -eps 5000. -pp 70 -r 1 .518 -kr 1.518 -ss kernels/JKernel_ss.fits -di kernels/JKernel_zi.fits -fi NOAA9 779_all_io_mean.fits -fo noaa9779_ss -kms -nc ttnoiseCovar_128128_r1.518.fits

Expected Output

apart from standard inversion output, a 4-dimensional file noaa9779_ss_invNoiseCovariance.fits will be created in the working directory. The diagonal elements of this file will represent the error estimates for slowness perturbation for each depth layer.

Case 4: Resolution kernels for Helioseismic Inversion

Description

Evaluate how well the slowness perturbation model estimates are localised

Input

mean travel times & configuration file
resampled slowness sensitivity kernels (Rytov approximation) JKernel_1.518.fits

./inv -ki kernels/JKernel_1.518.fits -eps 5000. -pp 70 -r 1 .518 -kr 1.518 -ss kernels/JKernel_ss.fits -di kernels/JKernel_zi.fits -fi NOAA9 779_all_io_mean.fits -fo noaa9779_ss -kms -ak

Expected Output

apart from standard inversion output, a 4-dimensional file noaa9779_ss_resolKernels.fits will be created in the working directory. The rows of the files are averaging kernels for the inversion, showing how well localised the model estimates are.

NOTE: All .fits files have been moved from the attachments table to http://msslxx.mssl.ucl.ac.uk:8080/eSDO/algorithms/LHI/LHI.html.

-- ElizabethAuden - 04 Aug 2006

Edit | Attach | Watch | Print version | History: r21 < r20 < r19 < r18 < r17 | Backlinks | Raw View | More topic actions
Topic revision: r21 - 2008-01-14 - ElizabethAuden
 
This site is powered by the TWiki collaboration platform Powered by PerlCopyright © 2008-2018 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback