HMI Synoptic Maps Produced by NSO/NISP
Anna L. H. Hughes, Luca Bertello, Andrew R. Marble, Niles A. Oien, Gordon Petrie, Alexei A. Pevtsov
HHMI Synoptic Maps Produced by NSO/NISP
Anna L. H. Hughes, Luca Bertello, Andrew R. Marble, Niles A. Oien,Gordon Petrie, and Alexei A. PevtsovNational Solar ObservatoryMay 12, 2016
Technical Report No.
NSO/NISP-2016-002
Abstract
Recently, the National Solar Observatory (NSO) Solar-atmosphere Pipeline Working Group(PWG) has undertaken the production of synoptic maps from Helioseismic and Magnetic Imager(HMI) magnetograms. A set of maps has been processed spanning the data available for 2010-2015 using twice daily images (taken at UT midnight and noon) and running them throughthe same algorithms used to produce SOLIS/VSM 6302l mean-magnetic and spatial-variancemaps. The contents of this document provide an overview of what these maps look like, andthe processing steps used to generate them from the original HMI input data. a r X i v : . [ a s t r o - ph . I M ] M a y MI Synoptic Maps Produced by NSO/NISP 2
Contents
The goal of this project has been to create a series of integral magnetic synoptic maps using HMIdata (Schou, et al., 2012) and run using the same algorithms as those that produce the spatial-variance synoptic maps outlined in Bertello, et al. (2014) and derived from NSO SOLIS (SynopticOptical Long-term Investigations of the Sun) VSM (Vector Spectromagnetograph) 630.2 nm data.The HMI synoptic maps that we have produced are all integral synoptic maps (spanning 0-360 ◦ of a single Carrington rotation), in Carrington-longitude–by–sine(latitude) binning, and inNSO-low-resolution format (360x180 map bins, where contributing observations are weighted bycosine (∆longitude) relative to the observed longitude of central meridian).Unlike the SOLIS/VSM spatial-variance maps produced to date, the HMI synoptic maps havebeen run using both longitudinal- and vector-observed magnetograms, where the final data-productsfor each are detailed in § § For the HMI-longitudinal synoptic maps, we have used data from the HMI m 720s series (longitu-dinal magnetograms covering a 12-minute integration window, Scherrer, et al. (2012)). In order toproduce maps of radial magnetic flux, we have projected the line-of-sight flux values into pseudo-radial values using the assumption of a perfectly radial magnetic field at the photosphere: B r, pseudo = B LOS / cos ( ρ ) , (1)where B LOS is the observed LOS flux, and ρ is the center-to-limb (or heliocentric) angle betweenthe line-of-sight vector and the local vertical.Additionally, with the reasonable levels of quiet-sun sensitivity provided by longitudinal obser-vations, basic methods for filling in unobservable or poorly observed polar fields become viable.MI Synoptic Maps Produced by NSO/NISP 3Figure 1: Set of example image frames for an HMI-LOS-derived synoptic map: Carrington rotation2145 centered, on April 1 st , 2014.Therefore, for the HMI-LOS derived maps, we have provided a pole-filled version of the mean–pseudo-radial–flux map as an additional frame. Some methods of pole-filling interpolate spatiallyand temporally across the pole from well-observed dates/latitudes (Sun, et al., 2011). In our case,the polar fields are filled in using a cubic-polynomial surface fit to the currently observed fields atneighboring latitudes. The fit is performed on a polar-projection of the map using low standard-deviation-to-fit measurements only, and the high-latitude fit is then integrated into the observedsynoptic map, weighting toward the pole.A set of example maps derived from HMI-longitudinal magnetograms is shown in Figure 1,while the file-name structure that we have used and the specifics of the FITS-file frame contentsare outlined below. Filename Structure: ‘xbx73
YYMMDD t HHMM c CCCC
000 int-err dim-180 source-SDO-HMI.fits.gz’ ◦ ‘xbx73’: This is the product code that denotes HMI synoptic maps derived from photosphericlongitudinal magnetograms. ◦ ‘ YYMMDD t HHMM ’: This is the time-stamp assigned to the map. For Integral synopticmaps, NSO uses the date and time corresponding to the midpoint of a given Carringtonrotation. ◦ ‘c CCCC
FITS-frame Contents:frame units title / description
Weighted-mean Radial Flux Density:
The mean value of the radially-projected HMI-LOS magnetogramsfor each longitude-sine(latitude) map bin. Each input observation is spa-tially weighted to emphasize contributions observed near the central merid-ian.2 Gauss
Spatial RMS Estimate:
The weighted statistical variance of all radially-projected HMI-LOS-magnetogram-values contributing to a given longitude-sine(latitude) bin(corresponding to the mean-flux values in Frame 1).3 counts
Sum-of-Weights:
The sum of weights into each map bin. This includes both the∆longitude–versus–central-meridian weighting applied across each inputobservation, as well as the count of sky-image pixels contributing to eachobserved longitude-sine(latitude) bin.4 Gauss
Pole-filled Mean Radial Flux Density:
The pole-filled version of Frame 1.
For the HMI-vector synoptic maps, we have used data from the HMI b 720s series (fully disam-biguated vector magnetograms, Hoeksema, et al. (2014)), choosing to apply the results of theRadial-accute disambiguation for the regions of quiet sun (Metcalf, et al., 2006; Leka, et al., 2009).As this is vector data, the mean-radial-flux map in these files is for true-radial flux. Addition-ally, we have mapped the values for the mean poloidal and toroidal fluxes, and computed thespatial-variance of these quantities. For these additional-component maps, we have used the samecosine (∆longitude) weighting that broadly emphasizes fluxes observed near central meridian.A set of example maps derived from the HMI-vector magnetograms is shown in Figure 2, whilethe file-name structure that we have used and the specifics of the FITS-file frame contents areoutlined below. Filename Structure: ‘xbx93
YYMMDD t HHMM c CCC
000 int-err dim-180 source-SDO-HMI.fits.gz’ ◦ ‘xbx93’: This is the product code that denotes HMI synoptic maps derived from photosphericvector magnetograms. ◦ ‘ YYMMDD t HHMM ’: As in § ◦ ‘c CCCC § st , 2014.MI Synoptic Maps Produced by NSO/NISP 6 FITS-frame Contents:frame units title / description
Weighted-mean Radial Flux Density:
The mean value of the radial (outward) flux (measured from HMIvector magnetograms) for each longitude-sine(latitude) bin. Each inputobservation is spatially weighted to emphasize contributions observed nearthe central meridian.2 Gauss
Weighted-mean Poloidal Flux Density:
The mean value of the poloidal (southward) flux (measured from HMIvector magnetograms) for each longitude-sine(latitude) bin. Each inputobservation is spatially weighted to emphasize contributions observed nearthe central meridian.3 Gauss
Weighted-mean Toroidal Flux Density:
The mean value of the toroidal (+longitude-ward) flux (measuredfrom HMI vector magnetograms) for each longitude-sine(latitude) bin.Each input observation is spatially weighted to emphasize contributionsobserved near the central meridian.4 Gauss
Radial-flux Spatial RMS Estimate:
The weighted statistical variance of all HMI-vector radial-fluxmagnetogram-values contributing to a given longitude-sine(latitude) bin(corresponding to the mean-flux values in Frame 1).5 Gauss
Poloidal-flux Spatial RMS Estimate:
The weighted statistical variance of all HMI-vector poloidal-fluxmagnetogram-values contributing to a given longitude-sine(latitude) bin(corresponding to the mean-flux values in Frame 2).6 Gauss
Toroidal-flux Spatial RMS Estimate:
The weighted statistical variance of all HMI-vector toroidal-fluxmagnetogram-values contributing to a given longitude-sine(latitude) bin(corresponding to the mean-flux values in Frame 3).7 counts
Sum-of-Weights:
The sum of weights into each map bin. This includes both the∆longitude–versus–central-meridian weighting applied across each inputobservation, as well as the count of sky-image pixels contributing to eachobserved longitude-sine(latitude) bin.
The following sub-sections provide a basic map of the various processing stages required to ingestHMI sky-image magnetograms and output SOLIS–spatial-variance–style synoptic maps. TheseMI Synoptic Maps Produced by NSO/NISP 7stages include:1. Ingest and prep of HMI sky images ( § § § In order to prepare the HMI magnetograms for heliographic and synoptic mapping, a few thingsneed to happen, including: download the magnetograms from the Joint Science Operations Center(JSOC) site, update the image orientation and a few FITS-header keywords to comply with SOLIS-pipeline expectations, and — in the case of the vector magnetograms — calculate the heliographicmagnetic-vector components from the HMI input frames. The layout of the code calls looks likethis:1. Call backfillMagnetograms.sh
N1 N2 :- For each day N1 to N2 days ago, requests a download of the UT 00:00 and UT 12:00magnetograms from JSOC and places the results in an NSO-accessible data-keep direc-tory.2. Call hmi serrmaps intake2fits BatchRun.sh [-v] START STOP OUTDIR :- For each day from
START to STOP : • Searches for available downloaded JSOC files,
INFILE s. • Calls hmi serrmaps intake2fits [-v]
INFILE OUTDIR → Opens the rice-compressed
INFILE file. → Rotates the FITS image by 180 ◦ to place Solar-north at the top. → Re-writes the FITS header using SOLIS-style sectioning. → Adds (primarily duplicate) keywords to the FITS header to account for theupdated image geometry and allow for data read-in by SOLIS downstream pro-cessing. → Outputs the results to a gzipped file placed in a data keep within
OUTDIR andusing the NSO-style file-naming conventions.3.
IF(vector):
Call hmi serrmaps intakevbundle BatchRun.sh
START STOP :- For each day from
START to STOP : • Searches for available ingested gzipped FITS files with flavor tag ‘ type-b-720s-field’,
FIELDFILE s. • In IDL, calls hmi serrmaps intakevbundle , FIELDFILE : → Using input
FIELDFILE filename to extrapolate, reads in the full file set nec-essary (‘-field’, ‘-inclination’, ‘-azimuth’, ‘-disambig’) to compute magnetic-fluxvector components. → Applies the disambiguation results (Radial-accute in the quiet sun) to the az-imuth image by adding 180 ◦ to all pixels where disambig is true. → Calls chcoord3.pro to define the heliographic coordinates of each image pixel.MI Synoptic Maps Produced by NSO/NISP 8Figure 3: Example image for an HMI-LOS magnetogram taken Dec. 1 st , 2013 at 00:00 UT. → Computes the line-of-site and transverse magnetic-vector components, then ro-tates them into the local-surface heliographic plane(s). → Outputs the three frames of heliographic vector components into a FITS filewith the flavor tag ‘ type-b-720s-helio’.
LOS magnetograms:
For the LOS magnetograms, an example output image (from step 2) isshown in Figure 3. These FITS files have only a single image frame, containing the LOS magneticflux measured by HMI. They are given file names with the structure:‘x4x72
YYMMDD t HHMMSS source-SDO-HMI type-m-720s.fits.gz’ ◦ ‘x4x72’: This is the product code that denotes HMI sky images of photospheric longitudinalmagnetograms. ◦ ‘ YYMMDD t HHMMSS ’: This is the observation’s time-stamp. ◦ ‘ type-m-720s’: This indicates HMI–line-of-sight–magnetogram source data, regardless of im-age type. Vector magnetograms:
For the vector magnetograms, an example file set of ingested (outputfrom step 2) data are shown in Figure 4. In the ‘-azimuth’ file, angles are measured from the +yimage axis and increase counter-clockwise. In the ‘-disambig’ file, true values for the Radial-acutedisambiguation are indicated with integer values 4,5,6 and 7 (for Random disambiguation: 2,3,6,7;for Potential-acute disambiguation: 1,3,5,7).The frames for the corresponding heliocentric–magnetic-vector–components file (output fromstep 3) are shown in Figure 5. These final-vector sky-image output FITS files have naming struc-tures and frame contents as outlined below.
Filename Structure: ‘x4x92
YYMMDD t HHMMSS source-SDO-HMI type-b-720s-helio.fits.gz’ ◦ ‘x4x92’: This is the product code that denotes HMI sky images of photospheric vector mag-netograms. ◦ ‘ YYMMDD t HHMMSS ’: This is the observation’s time-stamp.MI Synoptic Maps Produced by NSO/NISP 9Figure 4: Set of example input image frames for an HMI-Vector magnetogram taken Oct. 18 th ,2015 at 00:00 UT.Figure 5: Set of example image frames for an HMI-Vector-derived Heliocentric-vector magnetogramobserved Oct. 18 th , 2015 at 00:00 UT. ◦ ‘ type-b-720s-helio’: Regardless of image type, this indicates HMI-vector-magnetogram sourcedata mapped into heliocentric vector components. FITS-frame Contents:frame units title / description
Radial flux (outward):
HMI b 720s magnetogram radial-flux vector component.2 Gauss
Poloidal flux (southward):
HMI b 720s magnetogram poloidal-flux vector component.3 Gauss
Toroidal flux (+longitude-ward):
HMI b 720s magnetogram toroidal-flux vector component.MI Synoptic Maps Produced by NSO/NISP 10
Once the HMI sky images have been prepared for ingest into SOLIS synoptic-map processing ( § hmi serrmaps remap BatchRun.sh [-v] START STOP OUTDIR :- For each day from
START to STOP : • Searches for available prepped sky images,
SKYFILE s. • In IDL, calls hmi serrmaps remap , SKYFILE , OUTDIR , /tokeep, /sinlat: → Reads in the
SKYFILE image frame(s). → Calls chcoord3.pro to define the heliographic coordinates of each image pixel,and for all four corners of each pixel. → IF(longitudinal):
Projects the line-of-sight flux values into purely radial fluxvalues (as per Equation 1). → Defines the Carrington-longitude bounds for the observation to set the bins forthe heliographic output map. → Sorts all on-disk image pixels into weighted longitude-sine(latitude) bins. Pixelsthat cover multiple heliographic bins may be broken up into as many as 25 (5x5)sub-pixels for heliographic binning. (
Note:
This matches the spatial resolutionof the SOLIS spatial-variance-map sub-pixel binning, where image pixels arebroken up into 10x10 sub-pixels but derive from observations of half the spatialresolution as HMI.) → For each heliographic bin, computes: ◦ the sum-of-weights (number of contributing pixels) ◦ the mean magnetic flux ◦ the RMS flux variance ◦ the sum of squared weights ◦ the mean of squared fluxes → Note:
For vector magnetograms , the mean, RMS, and mean-squaredfluxes are computed individually for all three vector components. → Outputs the resulting heliographic maps of computed quantities into a FITSfile using the NSO-style file-naming convention and placed in a data keep in
OUTDIR . Pseudo-radial Heliographic Maps:
An example of the frames output for a pseudo-radial heli-ographic map are shown in Figure 6. These FITS files have naming structures and frame contentsas outlined below.
Filename Structure: ‘x9x73
YYMMDD t HHMMSS map-err dim-180 source-SDO-HMI type-m-720s.fits.gz’ ◦ ‘x9x73’: This is the product code that denotes HMI heliographic remaps of photosphericlongitudinal data. ◦ ‘ YYMMDD t HHMMSS ’: This is the observation’s time-stamp.MI Synoptic Maps Produced by NSO/NISP 11Figure 6: Set of example image frames for an HMI-LOS-derived heliographic remap for an obser-vation taken Dec. 1 st , 2013 at 00:00 UT. ◦ ‘ type-m-720s’: Regardless of image type, this indicates HMI–line-of-sight–magnetogramsource data. FITS-frame Contents:frame units title / description
Weighted-mean Radial Flux Density:
Mean of radially-projected HMI-LOS magnetic flux at each helio-graphic bin.2 Gauss
Spatial RMS Estimate:
Statistical variance of all radially-projected HMI-LOS-flux values ateach heliographic bin.3 Gauss Mean squared-Radial Flux:
Mean of squared pseudo-radial flux values at each heliographic bin.4 counts
Sum-of-Weights:
Sum of weights (image-pixel fractions) into each heliographic bin.5 counts Sum-of-squared-Weights:
Sum of squared-weights into each heliographic bin.MI Synoptic Maps Produced by NSO/NISP 12Figure 7: Set of example image frames for an HMI-Vector-derived heliographic remap for an ob-servation taken Dec. 1 st , 2013 at 00:00 UT. Vector Heliographic Maps:
An example of the frames output for a vector heliographic map areshown in Figure 7. These FITS files have naming structures and frame contents as outlined below.
Filename Structure: ‘x9x93
YYMMDD t HHMMSS map-err dim-180 source-SDO-HMI type-b-720s-helio.fits.gz’ ◦ ‘x9x93’: This is the product code that denotes HMI heliographic maps of photospheric vectormagnetograms. ◦ ‘ YYMMDD t HHMMSS ’: This is the observation’s time-stamp. ◦ ‘ type-b-720s-helio’: Regardless of image type, this indicates HMI-vector-magnetogram sourcedata mapped into heliocentric vector components.MI Synoptic Maps Produced by NSO/NISP 13 FITS-frame Contents:frame units title / description
Mean Radial (outward) Flux Density:
Mean of HMI-vector radial flux at each heliographic-coordinate bin.2 Gauss
Mean Poloidal (southward) Flux Density:
Mean of HMI-vector poloidal flux at each heliographic-coordinate bin.3 Gauss
Mean Toroidal (+longitude-ward) Flux Density:
Mean of HMI-vector toroidal flux at each heliographic-coordinate bin.4 Gauss
Radial-flux Spatial RMS Estimate:
Statistical variance of all HMI-vector radial-flux values into eachheliographic-coordinate bin.5 Gauss
Poloidal-flux Spatial RMS Estimate:
Statistical variance of all HMI-vector poloidal-flux values into eachheliographic-coordinate bin.6 Gauss
Toroidal-flux Spatial RMS Estimate:
Statistical variance of all HMI-vector toroidal-flux values into eachheliographic-coordinate bin.7 Gauss Mean squared-Radial Flux:
Mean of squared radial-flux values at each heliographic-coordinatebin.8 Gauss Mean squared-Poloidal Flux:
Mean of squared poloidal-flux values at each heliographic-coordinatebin.9 Gauss Mean squared-Toroidal Flux:
Mean of squared toroidal-flux values at each heliographic-coordinatebin.10 counts
Sum-of-Weights:
Sum of weights (image-pixel fractions) into each heliographic-coordinate bin.11 counts Sum-of-squared-Weights:
Sum of squared-weights into each heliographic-coordinate bin.MI Synoptic Maps Produced by NSO/NISP 14
Once all of the heliographic remaps have been processed ( § ◦ of Carrington longitude, as follows:5. Call hmi serrmaps synoptic BatchRun.sh [-v] START STOP CARRFILE :- Uses
CARRFILE to look up the date ranges of the Carrington rotations,
CARRNUM s.- For each
CARRNUM ocuring between
START and
STOP : • Calls the IDL routine hmi serrmaps synoptic.pro for the specified
CARRNUM and data type (HMI-LOS or HMI-Vector): → Looks up the date range covered by
CARRNUM and searches the data keep fora list of all available heliographic remaps falling within that date range +/- anadditional 8 days. → Reads in the headers of the listed heliographic files in order to: * Define the range of longitude bins covered by each heliographic map. * Discard from the list any heliographic maps that fall entirely outside the0-360 ◦ longitude of CARRNUM (e.g., usually discards the maps from obser-vations taken 8 days before and after the Carrington-rotation date bounds). * Double-check various observation-quality keywords and discard any helio-graphic maps that fail. → For each heliographic-map file,
HRFILE , retained from the file list: * Reads in the
HRFILE image frames. * Rescales the values in the Weights frame by cosine (∆longitude) with respectto the central meridian. * Places all in-bounds heliographic-map data into the synoptic-map imagespace. In this step, each heliographicly mapped quantity for this observation(weights, fluxes, etc.) is saved into its own synoptic-map of an nfiles stackedset. → Once all of the heliographic maps have been loaded into the synoptic-map space,computes: ◦ the sum of weights in each synoptic-map bin ◦ the mean weighted-flux values in each synoptic-map bin ◦ the spatial variance of the flux values in each synoptic-map bin → Note a:
For vector maps , the mean-flux and spatial-variance values arecomputed individually for all three vector components. → Note b:
For any synoptic-map bin where the sum-of-weights equals 0, themean-flux value(s) is set to 0, and the spatial-variance(s) is flagged with thenonsense value -1000. → IF(HMI-longitudinal):
Calls hmi serrmaps polefiller sfit.pro to returna pole-filled version of the radial-flux map. → Outputs the final synoptic maps of computed quantities into a FITS file usingthe NSO-style file-naming convention outlined in § § Creation of synoptic maps from HMI magnetograms required a few choices as to the handling ofthe HMI data, and primary among them was which quiet-sun disambiguation results should beemployed to project the observed vector fields into heliographic coordinates.The -disambig file included with all HMI-vector magnetograms in the b 720s series providesthe HMI-disambiguation results as an image of true/false values answering whether the azimuthangle at a given pixel should be rotated by 180 ◦ relative to the value provided in the -azimuth file(Hoeksema, et al., 2014). For strong-field and near-strong-field pixels, the disambiguation is theresult of “annealing” using a minimum-energy algorithm. For weak-field pixels, the -disambig fileprovides results from three different disambiguation algorithms:1. A Potential-acute algorithm that works to align the field with a potential field extrapolatedfrom the vertical field component.2. A
Random disambiguation assignation.3. A
Radial-acute algorithm that selects the disambiguation that most closely aligns the fieldin the purely radial direction.The HMI documentation (JSOC Wiki - Disambiguation, 2014) recommends using
Acknowledgements
The authors would like to thank Yang Liu for assistance with the appropriate use of HMI FITS-fileheader keywords. This work was partially supported by NASA grant NNX15AN43G.
References
Bertello, L., Pevtsov, A. A., Petrie, G. J. D., and Keys, D. 2014. “Uncertainties in Solar SynopticMagnetic Flux Maps”, Solar Physics, , 2419-2431.Hoeksema, J. T., +11 co-authors 2014. “The Helioseismic and Magnetic Imager (HMI) VectorMagnetic Field Pipeline: Overview and Performance”, Solar Physics, , 3483-3530.“JSOC Wiki - Full-Disk Disambiguated Vector Magnetic Field from HMI.” http://jsoc.stanford.edu/jsocwiki/FullDiskDisamb . Joint Science Operations Center, May 2014, Web. 8 Apr. 2016.Leka, K. D., Barnes, G., Crouch, A. D., Metcalf, T. R., Gary, G. A., Jing, J., and Liu, Y. 2009.“Resolving the 180 ◦ Ambiguity in Solar Vector Magnetic Field Data: Evaluating the Effects ofNoise, Spatial Resolution, and Method Assumptions”, Solar Physics, , 83-108.Metcalf, T. R., +14 co-authors 2006. “An overview of existing algorithms for resolving the 180 ◦ ambiguity in vector magnetic fields: Quantitative tests with synthetic data”, Solar Physics, ,267-296.Scherrer, P. H., +12 co-authors 2012. “The Helioseismic and Magnetic Imager (HMI) Investigationfor the Solar Dynamics Observatory (SDO)”, Solar Physics, , 207-227.MI Synoptic Maps Produced by NSO/NISP 16Figure 8: For Carrington rotation 2161 centered around March 14 th Radial-acute (left column) versus
Random (right col-umn) weak-field disambiguation results. From top to bottom, the rows plot the mean-radial,-poloidal, and -toroidal flux, with a greyscale stretching from -20 to +20 Gauss.Schou, J., +20 co-authors 2012. “Design and Ground Calibration of the Helioseismic and MagneticImager (HMI) Instrument on the Solar Dynamics Observatory (SDO)”, Solar Physics, , 229-259.Sun, X., Liu, Y., Hoeksema, J. T., Hayashi, K., Zhao, X. 2011. “A New Method for Polar FieldInterpolation”, Solar Physics,270