HARP - HMI Active Region Patches

HARP Overview

A HARP (short for HMI Active Region Patch) is an enduring, coherent magnetic structure at the scale of a solar active region. The primary purpose of the HARP data series is to provide the practical geometric information needed to follow an evolving region as it crosses the solar disk. A HARP is initially identified in a sequence of HMI line-of-sight magnetograms. HARPs are typically observed over several days (possibly as long as a disk passage) and tracked from one image to the next. At each time step, the rectangular HARP bounding box is defined and a BITMAP that characterizes the pixels of the HARP is recorded.

Other summary parameters, such as total flux, are also determined. A more comprehensive data series, SHARPS, are descried at ["SpaceWeatherHARP"]'.

The definitive line-of-sight magnetic field HARPs are available in the data series hmi.Mharp_720s. The Mharp data series is indexed by two prime keys: an integer identifier, HARPNUM, and the time, T_REC. The HARP number is unique to the region during it's entire disk passage. A particular HARPNUM will often be linked with a NOAA Active Region (see the relevant keywords: NOAA_AR, NOAA_NUM, and NOAA_ARS). This data series provides pointers to information covering the entire disk passage for each HARP in the HMI catalog for each HARPNUM at each T_REC. Because we use 720s HMI data, some HARPs contain as many as 1500 distinct T_REC's, and some as few as three.

The HARPs are found by analyzing the active region masks in hmi.Marmask_720s (see ["ARmaskDataSeries"]). The masks are in turn derived from line-of-sight magnetograms, hmi.M_720s, and corresponding intensitygrams, hmi.Ic_noLimbDark_720s. All three of these data products are full-disk images in helioprojective-tangent coordinates (i.e., as projected on the focal plane in CCD coordinates, and not remapped to a latitude-longitude system). Links to the magnetogram and the masks are provided.

In addition to the keywords mentioned above, the HARP data series contains a 2D data segment, the BITMAP. Coded values in the BITMAP indicate exactly which pixels within the rectangular bounding box are part of the HARP at the time T_REC. The BITMAP is a rectangle with the size of the bounding box for the full-disk images referred to above, typically several hundred pixels in each dimension. The bounding box is larger than the HARP because it is sized to enclose the largest heliographic extent of the HARP during it's entire disk passage.

The observations can be found by following the included links to associated line-of-sight magnetograms.

For diagnostic or summary purposes plots of individual keywords, such as integrated flux or size of the HARP, are also useful. These can be made quickly and easily using the GRAPH capability of the JSOC lookdata program.

Methodology

Once localized sites that are magnetically active have been found (i.e., building on the full-disk active region masks) the HARP identification problem consists of two pieces: spatially grouping magnetic activity into objects on the scale of active regions, and tracking these objects from image to image. The grouping problem is harder, because flux emergence can cause formerly isolated ARs to merge. This means that a given HARP cannot be declared complete until it has disappeared from view or rotated off the visible disk. Consequently, final construction of the HARPs are delayed by about a month.

It is important to track HARPs up to the limb, so that all the history of the HARP can be taken in to account in making grouping decisions. Consequently, the grouping criterion takes the spherical geometry into account.

We expect it will be useful to have easy access to the precursors and successors of the HARP. So, we extrapolate the area containing the HARP backwards in time from where it was first detected, and forward from the time where it vanished, two days in each direction. (Or less, if the entire region would rotate off-disk in this time.) This has the effect of expanding the range of T_REC associated with each HARP.

The HARP identification component consists of two parts, a grouping/tracking component, implemented in Matlab, and a data ingestion component, implemented as a JSOC module.

Other Definitive HARP Data Series

HARPs are also used in the creation other data series when computation of the full disk observables is prohibitive. Currently inverted vector field data are available for selected times in hmi.ME_720s_fd10_HARP and hmi.ME_720s_e15w1332_HARP. The names are a little different because the format is different; these data sets are 4096**2 images like their namesake full disk counterparts, but at each time step only the data within all of the identified HARPs has been determined. Other data sets with disambiguated vector magnetic field data will follow.

Near Real Time HARP Data Series

Near real time HARP information is most useful for space weather forecasting. The need for rapid processing means that some convenient features of the definitive HARPs cannot be realized because the future of regions is not known and we cannot afford to go back in time to adjust previous times step. For example, the heliographic size of the HARP will change as it evolves; HARPs representing multiple emergence sites may merge; ancillary information, such as NOAA AR numbers, may not be known; and data before the first emergence of flux will not be included. The numbering of HARPs and HARP_nrt's will differ. Nevertheless, since we cannot afford to reprocess much of the full disk vector field data, concentrating only on the interesting active regions allows us to maintain a rapid cadence.

The Mharp data are generally available shortly after the data are collected and look much like the definitive HARPs, except that when regions merge there will be discontinuities in some summary quantities. A movie of recent Mharps can be found at [http://jsoc.stanford.edu/data/hmi/movies/latest/latestHARP.mp4 Latest MHarp MP4].

The Milne-Eddington inversion data (and later the disambiguated vector B data) take longer to produce and currently lag behind. These data sets are currently like their partially filled-in full-disk counterparts. Space-weather HARPs will be computed from the regularly disambiguated data as soon as it becomes available.

HARP Feature Scale

The following images show how the convolution kernel used for spatial grouping of active regions compares to a small active region at disk center. The units on the top two plots are HMI pixels. The bottom plot shows the convolution kernel at the same scale.

[http://sun.stanford.edu/~turmon/jsoc-wiki/kernel-scale-ar-mag-small.png] [http://sun.stanford.edu/~turmon/jsoc-wiki/kernel-scale-ar-mask-small.png]

[http://sun.stanford.edu/~turmon/jsoc-wiki/ar-grouping-kernel-center-small.png]

When at the limb, the convolution kernel is foreshortened as shown at right

[http://sun.stanford.edu/~turmon/jsoc-wiki/ar-grouping-kernel-limb-small.png]

Example

The following images are approximate data segments extracted from the large February 2011 active region. The corresponding HARP contains hundreds of T_REC values; we show only five. The orange blob outlines the contents of the HARP. The black pixels, most of which are inside the HARP, are all pixels declared active in the mask. Some clumps of active pixels are not large enough to constitute a HARP.

The white rectangle is not part of the data segment. It is the bounding box in pixel coordinates which contains the entire HARP, and its coordinates are part of the keywords. There is a corresponding bounding box, not shown, in Stonyhurst latitude-longitude coordinates, which is also recorded in the keywords, as MINLON0, MINLAT0, etc.

[http://sun.stanford.edu/~turmon/jsoc-wiki/track-movie-2011-feb-066-zoom.png] [http://sun.stanford.edu/~turmon/jsoc-wiki/track-movie-2011-feb-126-zoom.png] [http://sun.stanford.edu/~turmon/jsoc-wiki/track-movie-2011-feb-186-zoom.png] [http://sun.stanford.edu/~turmon/jsoc-wiki/track-movie-2011-feb-246-zoom.png] [http://sun.stanford.edu/~turmon/jsoc-wiki/track-movie-2011-feb-306-zoom.png]

Anonther Example Showing HARPs on May 9, 2012

The image below is a snapshot showing the HARPs identified on the solar surface on May 9, 2012. Click images for details.

ImageLink(MovieFrame_HARP,MovieFrame_HARP,width=512,height=512)

Bitmap for HARP 1638, NOAA 11476

Cut-out of Full Disk Mask

Cut out of Full Disk Magnetogram

ImageLink(BitMap_HARP,BitMap_HARP,width=400,height=256)

ImageLink(BitMask_HARP,BitMask_HARP,width=400,height=245)

ImageLink(BitMag_HARP,BitMag_HARP,width=400,height=245)

Bitmap

Keywords

A complete description of the keywords is available in the SERIES CONTENT tab of the JSOC lookdata tool. Besides the standard HMI keywords (observation geometry, time, and WCS), we determine these keywords for HARP at each T_REC where it was observed:

Name

unit

Description

MINLON

degree

Minimum longitude for disk transit

MINLAT

degree

Minimum latitude for disk transit

MAXLON

degree

Maximum longitude for disk transit

MAXLAT

degree

Maximum latitude for disk transit

OMEGA

degree/day

Rotation rate

NPIX

none

Number of pixels within the identified region

SIZE

mH

Projected area of identified region on image in micro-hemisphere

AREA

mH

De-projected area of identified region on sphere in micro-hemisphere

NACR

none

Number of active pixels

SIZE_ACR

mH

Projected area of active pixels on image in micro-hemisphere

AREA_ACR

mH

De-projected area of active pixels on sphere in micro-hemisphere

MTOT

weber

Sum of absolute LoS flux within the identified region

MNET

weber

Net LoS flux within the identified region

MPOS_TOT

weber

Absolute value of total positive LoS flux

MNEG_TOT

weber

Absolute value of total negative LoS flux

MMEAN

gauss

Mean of LoS flux density

MSTDEV

gauss

Standard deviation of LoS flux density

MSKEW

none

Skewness of LoS flux density

MKURT

none

Kurtosis of LoS flux density

MINLAT0

degree

Minimum Stonyhurst latitude of pixels within the patch

MINLON0

degree

Minimum Stonyhurst longitude of pixels within the patch

MAXLAT0

degree

Maximum Stonyhurst latitude of pixels within the patch

MAXLON0

degree

Maximum Stonyhurst longitude of pixels within the patch

FWT_LAT

degree

Stonyhurst latitude of flux-weighted center of active pixels

FWT_LON

degree

Stonyhurst longitude of flux-weighted center of active pixels

FWTPOS_LAT

degree

Stonyhurst latitude of flux-weighted center of positive flux

FWTPOS_LON

degree

Stonyhurst longitude of flux-weighted center of positive flux

FWTNEG_LAT

degree

Stonyhurst latitude of flux-weighted center of negative flux

FWTNEG_LON

degree

Stonyhurst longitude of flux-weighted center of negative flux

T_FRST

TAI

T_REC of first frame of this HARPNUM

T_LAST

TAI

T_REC of last frame of this HARPNUM

Note