Time-Distance Pipeline for the Upper 30Mm Convection Zone

1
Time-Distance Pipeline for the Upper 30Mm
Convection Zone

• Status 11/5/2007

2
Time-Distance Helioseismology
Full-disk velocity, v(r,T,F), And sound speed,
cs(r,T,F), Maps (0-30Mm)
Filtergrams
Carrington synoptic v and cs maps (0-30Mm)
Doppler Velocity
Time-distance Cross-covariance function
Tracked Tiles Of Dopplergrams
Wave travel times
High-resolution v and cs maps (0-30Mm)
Deep-focus v and cs maps (0-200Mm)
3
Task
Output
Input
Active region structures and flows
Synoptic flow maps
Dopplergrams
4
Status Cross-Covariance
Task Do filtering and compute cross-covariance

• Status
• filtering code ready, comparisons among Tom,
  Sebastien and myself have achieved excellent
  results
• cross-covariance computation code ready,
  comparisons also done, and excellent results
  achieved
• selection of annulus pixels algorithm has
  been compared between Tom and me, a Fortran code
  needs to be developed for the active region
  cases. Synoptic flow maps will use pre-computed
  locations.

Summary Nearly done except one subroutine for
on-fly computation of annulus pixels. Module is
needed.
5
Status Travel Time Fitting
Task Fit cross-covariances to derive acoustic
travel times

• Status
• As of last Friday, comparisons among Tom,
  Sebastien and myself are still going on. I am
  optimistic that excellent results will be
  achieved.
• Codes in Fortran are largely ready. Only some
  fine-tuning is needed in some parameters.
• Fitting code based on Gizon-Birch method was
  written by Sebastien and ready to be incorporated
  into main code.

Summary Nearly done except incorporating
Sebastiens subroutine. Module is needed.
6
Status Sensitivity Kernels
Task Compute sensitivity kernels as input for
the inversion code

• Status
• Ray-approximation kernels a few codes exist,
  some comparisons have been planned to finalize
  this.
• Born-approximation Birch code ready for use
  with some opitmization that is being carried out
  by Konstantin. Also MPI can provide this.

Summary in a pretty good shape. Should be
finished in a couple of months.
7
Status Inversions
Task Develop code to do inversions

• Status
• Pipeline work on this part is barely started.
• Codes for two different algorithms (LSQR and
  MCD) exist. Probably both algorithms will be put
  into pipeline.
• Some comparisons between different codes and
  different algorithms are needed.

Summary Barely anything has been done. A lot of
efforts are needed in this. An estimate of 6
months(?) is required for this part of work.
8
Additional Issues

• Poor or no noise estimations in both
  time-distance measurements and inversions. This
  needs a great deal of research efforts to
  implement it. Not sure whether this part can be
  accomplished by the time of March 2009.
• Sebastien has code for inversions with noise
  covariance.
• All codes are based on MDI observations. Some
  parameter changes and more testing is needed for
  HMI observations.
• Co-I asks for more support.
• Overall, it is very optimistic to deliver all
  codes before March, 2009

9
Computer Requirement

• Now we are ambitious in computing 17 different
  annuli for HMI data
• Computing 17 annuli using 512-minute MDI
  full-disk data takes approximately1.5 hours on 1
  LWS cluster node for 1 of 9 regions selected on
  solar disk.
• We do not yet know whether we will use HMI full
  resolution to do this.
• We save fitted travel times and inverted 3D
  velocity and 1D sound-speed perturbation for
  approximately 15 depths. Every 8.5 hours, it
  takes 62MB saving space for MDI resolution.

10
Time-Distance Pipeline for Far-side Imaging

• Status 11/5/2007

11
Task
Output
Input
Far-side images
Dopplergrams
12
Status

• All codes for computing travel times are ready in
  Fortran. Some processing and making of maps after
  travel times are in IDL.
• Each far-side map is combined from one 4-skip map
  and one 5-skip map.
• Module is needed.
• Should be able to share a lot of things with
  holography far-side imaging, hence saving a lot
  of computing and storage resources.

Computer Requirement

• One map is made every 12 hours
• Using MDI medium-l data, one map needs about 1
  hour computing time using one LWS2 cluster node.
• Storage space is quite small.

13
Issues

• No additional support is needed.
• Though codes are ready, results are published,
  maps are not widely available online and seen by
  colleagues. Hence, it is useful to broadcast the
  results so as to receive feedbacks from
  colleagues.
• Statistical study regarding to how much we can
  believe in those images, which part of images is
  more reliable and which part is not reliable.
• Systematic comparison with holography far-side
  images is probably necessary.

14
Helioseismic Holography Pipeline

• Status 11/5/2007

15
Acoustic Holography
High-resolution v and cs maps (0-30Mm)
Filtergrams
Doppler Velocity
Tracked Tiles Of Dopplergrams
Egression and Ingression maps
Wave phase shift maps
Inversions
Far-side activity index
16
Status (I)
Task Compute egression and ingression
functions, and measure acoustic phase shifts by
cross-correlating egression and ingression
Status Codes are essentially existing for
MDI-type data, though some improvements are
necessary, and some tuning to fit HMI is also
required.
17
Status (II)
Task Compute sensitivity kernels, and invert
for flow fields based on phase-shift measurements
Status Still under development. Only very
limited experimental coding has been developed.
18
Status Far-side Imaging
Status Codes are ready and are already run in
MDI pipeline, though retooling to HMI is needed.
Researches on corrections, calibrations and
near-side limb imaging are carried out, but do
not yet promises anything for updating or
improving for HMI pipeline.
19
Issues

• Nearly all codes are needed to be re-adjusted to
  HMI.
• Without additional funding, the codes they
  provide will be quite limited in utility and
  scope
• use local plane-parallel geometry
• have limited or no corrections in
  large-scale artifacts
• no treatment on magnetic surface effect
• far-side images will have little or no
  calibration.