Predicting CMEs is of fundamental importance for space weather. Currently, we are unable to do so.
Since CMEs are magnetically driven, characterizing the evolution of the solar magnetic field using quantitative techniques is a useful component of the space weather community's effort to understand the physical processes leading to CMEs.
These data are intended to be used by solar physicists in studies of the evolution of the solar magnetic field. Hence, data are intended to be presented in user friendly, easily accessible formats, and analysis techniques are intended to be presented as clearly and transparently (to allow reproducibility) as possible.
Two data products are stored here,
two-dimensional local correlation tracking (LCT) flow maps based upon:
1)
calibrated MDI and 2)
preliminary (uncalibrated)
full disk magnetograms.
Insufficient computational resources prompted cessation of
automated tracking in early July 2006. We anticipate that additional
resources will be available shortly, and tracking should resume by
November 2006.
Maps are available for preliminary full disk magnetograms starting
from 01 Nov 2005 to 10 July 2006.
Maps are available for calibrated full disk calibrated
magnetograms starting from 01 Feb 2006 to 07 July 2006.
In addition, the deprojected, full-disk magnetograms used to generate the flow maps (see the "How" section below) are also available, as are coordinate arrays of latitude and longitude.
The IDL routine, read_flct_ascii.pro, which can read these data products into IDL is available on this server.
The LCT tracking code used here, FLCT, is available on this server, from http://solarmuri.ssl.berkeley.edu/overview/publicdownloads/software.html.
From June 2006 onwared, the Pipeline used vel_ccor12, while magnetograms from 04 May 2006 and before were tracked with vel_ccor10.
The codes used in the deprojection and other analyses are also available on this web server, in http://solarmuri.ssl.berkeley.edu/~welsch/public/data/Pipeline/Software/.
In the near future, we plan to extend MEP to include magnetograms from SOLIS and GONG.
In addition to LCT analysis, we anticipate integrating analyses from other techniques that analyze the evolution of the solar photospheric field, including:
The pipeline is automated via the linux cron command: a script is check for new magnetograms, via wget, on remote data servers.
If new magnetograms are available, they are downloaded, and a batch-mode IDL procedure (here is a link with useful info, but some errors) is run. In general, this procedure generates: run-time information, which is piped to a log file; graphics information, which is processed into PNG files through IDL's z-buffer; and numerical output, which is stored in gzipped ASCII files.
Next, the output files are transferred to this web server using a technique for secure, scriptable file transfer.
To deproject the magnetograms, pixels' heliocentric latitudes and longitudes were determined using the IDL procedure pix2helio.pro, and pixels' values were interpolated, via IDL's triangulation routines, onto a regular grid in Mercator coordinates.
The Mercator projection was chosen because it is a conformal mapping: this projection distorts length scales, but not directions. This means that one flow component determined from correlating the projected data is not distorted relative to the other. Both components of the flow, however, must be scaled, by a factor of cos(latitude).
In addition to the FLCT flow arrays, the Mercator projections of the magnetograms' inner 30o are saved in the ASCII data files.
Flows were calculated for all pixels within 30o (angular distance in combined latitude and longitude) of disk center with absolute field strengths above 10 "units" --- because these are uncalibrated magnetograms, the field strengths are not provided! (Mean absolute preliminary magnetogram pixel values, among nonzero pixels, are ~3 +/- 4G.)
The Gaussian windowing function's width was set to 10.6 pixels.
Pixel values were not corrected for projection effects (one option for doing so would be to make the "radial approximation," and divide by the cosine of the angular distance from disk center [i.e., divide by a number greater than > 0.866]).
We used the FLCT code (Welsch et al. 2004) to perform the local correlation tracking. The FLCT code used on data from 05 May 2006 onward was modified to employ a second-order Taylor series estimation to find the maximum of the cross-correlation function (as opposed to the cubic-convolution interpolation described in this paper).
See
pipeline_MDI_idl_batch.pro
For each tracked pair of magnetograms, the deprojected magnetograms and the vflow maps (one 2d-array for vx, one 2D-array for vy) are stored on this web server, in gzipped ASCII format, e.g.,
FLCT_MDI_2006_02_06_00:00:03_ascii.dat.gz.
The IDL routine, read_flct_ascii.pro, which can read gunzipped versions of these data profiles into IDL, is available on this server.
In addition, for each pair of tracked magnetograms, several graphics files are also stored on this web server, e.g.,
