Measuring the Vector Magnetic Field of AR 11158 with HMI

The first disambiguated vector field data released by the HMI project for science analysis is a 5-day interval of 12-minute active region patches encompassing AR 11158. The region developed rapidly from February 12-16, 2011 and produced the first X-class flare of Solar Cycle 24 at ~01:56 UT on February 15.

The vector magnetic field is derived from the standard HMI 720-second Stokes-parameter data series, hmi.S_720s. Patches on the Sun that include AR 11158 were identified using the HMI Active Region Patch (HARP) code and then enlarged to provide a surrounding buffer of weak field. The VFISV inversion code was run with very strict convergence criteria to determine the optimal values; this required additional computing resources compared to the expected standard pipeline processing. The disambiguation was computed with spherical geometry in a limited region with fairly generous thresholds on field strength and very gradual annealing; this also required significant additional computing resources.

The HMI team is working to document the reduction pipeline and the uncertainties in the data a processing. We are reasonably confident in the results for this time series and are releasing this first HMI data series for test purposes now. Rather than wait longer, the HMI vector team asks that you treat these preliminary data with care and not publish results until the documentation is complete. The data series may change and the descriptions provided below will definitely improve as we move toward publication planned for early December, 2011. We welcome your feedback on the data and documentation.

You are strongly encouraged to read the ReleaseNotes that provide details and helpful information.

For this specific release, data have been exported into two special formats - direct cut outs and remapped images.

• For each time in the "cut out" series there are 10 FITS files giving the three vector magnetic field quantities, computed error estimates for each quantity, the three pair-wise correlations between the field quantities (the square root of the covariances), and a Info Map, with information about the quality of the processing at each pixel. Maps of additional quantities and covariances can be provided on request.

• Cut outs of the original CCD data are provided in a tracked 650 * 600 CCD-pixel window. The field components are the total flux, the inclination of the field relative to the local line of sight, and the azimuth measured counter-clockwise from 'up' on the CCD. Keywords in each FITS file provide information, including the solar geometry. Note that solar north is down in the cut-outs.

• The format perhaps more useful for modeling gives remapped images computed with a Lambert cylindrical equal area projection centered on the tracked region on a 600 * 600 grid with a spacing of 0.03 degrees. The vector field components are heliographic Br, Btheta, and Bphi. Errors for each component and a quality estimate for the processing, both at the nearest original pixel, are provided, for a total of 7 files per time step.

Vector field data must be used with care. There are several issues with vector magnetic field data that make inversions fail in some pixels and make definitive disambiguation problematic. Additionally, various systematic errors must be considered when using the data. These include daily, 12-hour, and other long-period variations associated with SDO's orbital velocity that cause large-scale spatial and temporal variations in sensitivity and in cross-talk in the Q and U components of the Stokes vector; these arise in part due to limitations in the knowledge of instrument characteristics. Various confidence and error estimates are provided, but they do not always indicate which pixels are suspect. To summarize:

• Inverted full disk vector magnetic field data show temporal and spatial variations. The temporal variation is probably due to the 24-hour variation of the orbital velocity that varies from -3.2 km/s to 3.2 km/s. The inverted field strength shows 24-hour, 12-hour and 6-hour periodicities in both strong-field and weak-field. The variation due to this effect is about 6.6% on average.
• The large-scale spatial variation can be seen in the inverted field strength. This large-scale pattern changes temporally, and appears to have a 24-hour period, too. The amptitude of this large-scale variation is about ± 50 Gauss in field strength.
• Sometimes there are some bad pixels in sunspot's umbrae and penumbrae. Those pixels have field very different from their very adjacent pixels. For the umbrae's bad pixels, they could appear individually or in patch.
• Inversion also returns line-of-sight velocity for magnetized plasma. This line-of-sight velocity matches well with the corresponding Doppler velocity. The line-of-sight magnetic field, derived from the difference of Doppler velocities measured from LCP and RCP, on the other hand, agrees with that from inversion results in weak-field, but shows significant discrepancy in high-field: The inverted line-of-sight field is higher than that from Doppler velocities.

AR 11158 emerged quickly in a complex way, with three major centers of flux emergence. During the build-up to the flare, mixed-polarity regions intensified and strong horizontal flows were apparent. At the time of the flare significant long-lasting changes in the horizontal magnetic field occurred.

The following outlines the analysis of the vector field for this region; a paper based on this outline will be submitted to Solar Physics shortly. Sections will be updated as they become available. Pointers to the data and more detailed treatments of some aspects of the processing follow. We strongly advise that you become familiar with the strengths and weaknesses of the data outlined below. Please contact a member of the VectorFieldTeam if you have questions about the data.

Measuring the Vector Magnetic Field of AR 11158 with HMI

• 1 Intro – Vector magnetic fields
  • 1.1 HMI Vector Field Overview
  • 1.2 AR 11158 – X-class Flare
• 2 Instrument and Data Processing
  • 2.1 HMI instrument
  • 2.2 Framelist for vector field
  • 2.3 Relevant calibration of HMI instrument
  • 2.4 Specifics of vector data camera, filtergrams
  • 2.5 Processing of filtergrams – dark, flat, cosmic rays, gap filling, distortion
  • 2.6 Pipeline and special HMI processing

• 3 IQUV calculation Section 3

  • 3.1 Computation of IQUV
  • 3.2 720 calculation – interpolation in time and space, averaging
  • 3.3 Other 720s quantities (very briefly)

• 4 VFISV inversion Section 4

  • 4.1 ME method and VFISV summary
  • 4.2 Specifics of implemented VFISV for this data series
    • 4.2.1 Weighting
    • 4.2.2 Filter Profile Approximation - Wavelength Hack
    • 4.2.3 Model Initialization
    • 4.2.4 Convergence Criteria, multiple minima
  • 4.3 Formal Error estimation

• 5 Disambiguation Section 5

  • 5.1 General method, issues, and uncertainties - planar and spherical
    • 5.1.1 Selection of Pixels to Anneal
    • 5.1.2 Potential Field Calculation
    • 5.1.3 Minimum Energy Methods
    • 5.1.4 Treatment of Weak Field Areas
  • 5.2 Computation Parameters

• 6 HMI.B Data series and access VectorPaper6

  • 6.1 Organization of data series
  • 6.2 Details of HMI.B contents
  • 6.3 Description of HARPs
  • 6.4 Export – general

  • 6.5 HMI Vector Data Versions and Referencing Data in publications - VectorDataReference

• 7 Projections - Section 7

  • 7.1 Native format – CCD coordinates etc.

  • 7.2 Details of Remapping and Vector Transformation on Export

    • 7.2.1 Mapping of pixels to regular coordinate system - Native CCD Cut-outs and Remapped CEA Coordinates

    • 7.2.2 Vector field components/directions - Transformation of Vector Components when Projected

    • 7.2.3 Handling of computed error quantities

• 8 Known and unresolved issues with HMI data and processing Section 8 Topics {10.26.2011}

  • 8.1 Periodic Errors
  • 8.2 Large-scale Variations
  • 8.3 Bad Pixels
  • 8.4 Comparison of Vector and Line-of-Sight Magnetic Field
  • 8.X Inversion Dependence on Initial Guess
  • 8.X Residual Cross-talk
  • 8.X Disambiguation issues
  • 8.X Other systematic and instrumental issues that affect data
• 9 Discussion of HMI measurements of AR 11158 - Validation
  • 9.1 General comments on determination of vector magnetic field
  • 9.2 Time Evolution of AR 11158 – features both real and otherwise
  • 9.3 Validation and limitations of HMI processing and data
  • 9.4 Qualitative comparison with B_LoS, Hinode, and other field measures
• 10 Summary
  • 10.1 Remaining work for pipeline modules
  • 10.2 Uses of data – NLFFF, DAVE4VM, other

Access to AR 11158 Vector Field Data

• See ReleaseNotes for AR 11158 Vector Field Data available in tar files.

• ReleaseNotes2 provides information about AR 11158 and additional regions processed in the same way and available through the standard JSOC facilitiy.

Movies of the Vector Field

• Br Cutout		Vector Field
  Vector Hedgehog	WEST	EAST	Vector Hedgehog (large)

  Intensity + Blos

  Full Disk Regions 1-16 Feb (HARP 0883)

  Full Disk Regions 16-28 Feb

Vector Field Data - One-Day Tar Files of 12-minute Cadence

• Data are currently available for scientific use.

  • ClickHereForDataRelease

One-Per-Day Sample Images and FITS Files

• Data are currently available for scientific use.

  • ClickHereForDataRelease

Other HMI data are available

A range of other HMI data pipeline products are available for this time period through the JSOC. These include the original Stokes parameters (the data series hmi.S_720s), the line-of-sight magnetic field (hmi.M), Doppler velocity (hmi.V), continuum intensity (hmi.Ic), line depth (hmi.Ld), and line width (hmi.Lw).

A longer synoptic time series of inverted (but not disambiguated) data are described at ME720se15w1332

SHARP data are available

The SHARP, or Spaceweather HMI Active Region Patch, data set are a collection of FITS header keywords with the HARP series included as DRMS linked segments. In the HARP data set, the vector magnetic field is determined for each patch and is available as a time series of FITS files. Many spaceweather quantities are calculated per patch. These quantities are known as SHARPs.

A collection of topical discussions of the processing

Details of some aspects of the processing and some older documentaiton can be found below. Watch for updates:

• IQUV Computation {updated 9/2011}

• VFISV Most complete {updated 10/2011}

  • VFISV Inversion Method Overview {updated 7/2011},

  • Weighting Scheme for IQUV in VFISV Inversion {updated 8/2011}

  • Filter / Line Profile Approximation (''the Hack'') {updated 8/2011}

  • hmi.ME inverted data detailed description {updated 7/2011}

  • hmi.B Info MAP - detailed description {updated 12/2011}

• Disambiguation Overview {updated 8/2011}

• Disambiguation Detail (incomplete) {updated 8/2011}

  • Notes on Determining Reliability of Disambiguation {GB - 8/2011}

• Systemic Effects {updated 10/2011}

  • A rough collection of Vector Field Systematic Issues {updated 7/2011}

• Remapping & Coordinate Transformation {updated 9/2011}

• Native CCD Cut-outs and Remapped CEA Coordinates

  • Note on Transformation of Vector Components when Projected

• Vector Field Documentation Directory {updated 10/2011}

• HMI Active Region Patches

  • HARP Data Series in JSOC/DRMS HARPDataSeries

  • HARP Quick-look Movies Tracked Movie

  • HARP Movie Format, Overlays, Annotations Movie Annotations

• Back to VectorMagneticField

• Back to HMIModuleTable

• Back to FrontPage