HARPs are active region patches identified in the HMI data and tracked during their entire visible lifetime. They may be associated with NOAA active regions, but there are many more HARPs.
The HMI pipeline analysis code automatically detects each active pixel in photospheric line-of-sight magnetograms and photograms. This code then identifies each patch on the full-disk imagery that encompasses an active region, and tracks every active region as it crosses the face of the solar disk. The HARP data series provides the geometrical information to follow the evolving region across the disk.
The SHARP data series provides not only the geometric information, but also extracts cut-outs of the HMI observable maps and provides summary space weather indices for the region at each time step.
HARPs are constructed in a two-stage process. For each 720s time step, the line-of-sight magnetic field and continuum intensity images are first analyzed to identify active pixels that might be part of an active region. This stage generates a full disk magnetically active pixel mask (Marmask) characterizing each pixel.
The second stage analyzes the time series of masks and groups the active pixels into HMI Active Region Patches (HARPs) that are tracked from one time step to the next. For each region at each time step, the geometry of the bounding box is provided and a bitmap characterizes the pixels within the box.
Once a region's vector magnetic field has been determined space weather quantities can be determined. The Space-weather HARPs (SHARPs) collect the cut outs of all of the interesting observables and derived indices. The SHARPs are presented in both native CCD and projected heliographic coordinates.
There are two kinds of HARPs, definitive and nrt. nrt data are processed as soon as data are available, typically 15 minutes for HARPs and a couple hours for the vector SHARPs. The nrt vector field is computed with faster algorithms and the HARP boundaries are constructed with data available at the time, so the quality is less good and the HARP regions sometimes merge and grow.
Definitive HARPs use the definitive boundaries that enclose at all time steps the maximal heliographic extent of the region attained during the entire disk passage. The entire disk passage is considered and regions that emerge or merge are linked for the entire disk passage. The HARP Number is not the same for the definitive and nrt SHARPs.
Additional information can be found at HARPS in the wiki.
SHARPs contain maps of vector and line-of-sight magnetic field, velocity, and intensity data for each HARP at each time step and include Keywords with space weather indices computed for the evolving, tracked regions. Fifteen space weather quantities useful for prediction -- such as emerging flux, field gradients, current, helicity, free energy, and more -- are calculated for each patch at each time step. Information can be found here or in more detail here.
The SHARP data series provide native CCD coordinate cut-outs of the complete set of inversion parameters: maps of the vector field, plasma parameters, uncertainties, and covariances, as well as the scalar observables: line-of-sight magnetic field, Doppler velocity, and continuum intensity.
The SHARP_CEA data series have been converted to heliographic coordinates and the vector quantities have been converted to a three-component vector. Only selected quantities are provided in the cylidrical equal-area projection data series.
The MHARP data series give geometric and other information about the HARP at each time step in the Keywords. In addition a BITMAP of the rectangular HARP identifies the blob of CCD pixels that are considered to be part of the active region and which of the blob pixels are 'active', i.e. above a threshold. Links to the full disk magnetogram and Marmask (see below) are provided.
Each HARP Number is associated with a time series of bitmaps definining a (perhaps complex) region as it transits the solar disk. The geometry of the definitive HARP is not finalized until the disk passage is complete. (This is different for the nrt HARPs). The largest heliographic extent of the region is tracked across the entire disk (or within two days of its emergence or disappearance). In addition to the geometric information, basic summary quantities about the region are reported, e.g. the total flux, the size of the region, any associated NOAA active regions, etc. The Mharp data series do not extract cut-outs of the observables (see SHARPs for that).
The nrt HARPs are computed as data become available each 12 minutes and are not able to look forward in time to anticipate future evolution. There is some persistence based on the history of the region, but the size and shape of the region can change significantly and multiple regions can merge over time.
The Mharp data series is indexed by two prime keys: an integer identifier, HARPNUM, and the time, T_REC. The numbers of nrt and definitive HARPs do not correspond.
The definitive Mharps are computed since the beginning of the HMI mission. The geometry of the rectangular Mharp bounding box is available in the keywords and a bitmap shows which blob of pixels are part of the HARP at each time step.
The nrt HARPs are the same, except the bounding box does not depend on knowledge of 'future' evolution of the region.
The Marmask data series are really intermediate products that provide at each 720s time step a magnetically active pixel map of the full disk. The pixels are identified in the 720s magnetograms and continuum images. No attempt to group the pixels or differentiate kinds of activity is currently implemented. See ARmaskDataSeries for more details.