HMI Instrument Yang Liu and HMI Team Stanford

HMI: Instrument Yang Liu and HMI Team Stanford University and Other Places yliu@sun. stanford. edu Page

Outline • HMI Instrument • Observables • Instrument Performance SDO SUMMER SCHOOL, 2010 Page 2

Instrument Overview – Optical Path Image Stabilization Mirror Beam Control Lens ¼ Waveplate ½ Waveplates Aperture Stop Blocking Filter Wideband Michelson Telescope lens set Telecentric Lens Lyot Polarizer Tuning Waveplates Narrowband Michelson ISS Beamsplitter and Limb Tracker Assembly BDS Beamsplitter Relay Lens Set Calibration lenses and Focus Blocks Front Window Filter CCD Shutter Assemblies CCD Fold Mirror Optical Characteristics: Focal Length: 495 cm Focal Ration: f/35. 2 Resolution: 1” Re-imaging Lens Magnification: 2 Focus Adjustment Range: 16 steps CCD Fold Mirror Filter Characteristics: Central Wave Length: 617. 3 nm Fe. I Front Window Rejects 99% Solar Heat Load Bandwidth: 0. 0076 nm Tunable Range: 0. 05 nm Free Spectral Range: 0. 0688 nm

Instrument Overview – HMI Optics Package (HOP) Connector Panel Z Focal Plane B/S Fold Mirror Shutters X Alignment Mech Limb Sensor Y Oven Structure Detector Michelson Interf. Lyot Filter CEBs Detector Vents Limb B/S Front Window Active Mirror Polarization Selector Focus/Calibration Wheels OP Structure Front Door Telescope Support Legs (6) Mechanical Characteristics: Box: 0. 84 x 0. 55 x 0. 16 m Over All: 1. 19 x 0. 83 x 0. 29 m Mass: 39. 25 kg First Mode: 63 Hz

Instrument Overview • Optics Package – Telescope section – Polarization selectors – 3 rotating waveplates for redundancy – Focus blocks – Image stabilization system – 5 element Lyot filter. One element tuned by rotating waveplate – 2 Michelson interferometers. Tunable with 2 waveplates and 1 polarizer for redundancy – Reimaging optics and beam distribution system – Shutters – 2 functionally identical CCD cameras • Electronics package • Cable harness

Polarization selector waveplates • Designed for installation one each into three Hollow Core Motors – Two 1/2 -wave plates and one 1/4 -wave plate • Specifications – Material: Single crystal quartz, grade A – Diameter 45 mm, clear aperture 34 mm, scratch/dig < 60 -40 within clear aperture – Surfaces flat to < /20, wedge < 2 arcmin, optical power < 1 wave – Optic axis parallel to face within 2 arcmin – Transmitted wavefront < /10 P-V – Retardances 10. 25 and 10. 50 +/ 0. 01 wave at 617. 3 nm at 20 C • Order status – Selected vendor: High Plains Optics – Delivered October 2004 – Parts are being measured at HAO in Boulder Polarization selector

Optical Path – Front Part Polarization selectors ISS mirror 1/2 1/4 1/2 Polarizing beamsplitter SDO SUMMER SCHOOL, 2010 Page 7 of 587

Polarization Selector Design • Rotating waveplates used – Hollow core motors identical to those used for tuning – Very similar to those used in MDI – Highly repeatable, <10” rms in rotation angle – 240 positions • Polarizing beamsplitter (and polarizer) to select desired linear component – Rejected light used for limb tracker • Desirable features of polarization selector: – All polarization states available (to within motor step size) – Redundancy – full or partial • At least 3 motors needed! – Clean LCP and RCP available – Simple retardances • ½, ¼, ½ design achieves all 4! (So does ¼, ¼, ½) – Any one motor can be stuck in any position and all states are still available – Keeping middle motor fixed gives lowest total wear SDO SUMMER SCHOOL, 2010 Page 8 of 587

FILTER: Summary of filter FWHMs, in order of light traveling • Front window: 50 A • Blocking filter: 8 A • Lyot element #2: 690 m. A • Lyot element #3: 1379 m. A • Beam Control Lyot element #4: 2758 m. A Lens • Lyot element #5: Wideband 5516 m. A Image Stabilization Mirror • Lyot element #1: 344 m. A (tuned) • Polarizer 172 m. A (tuned) Wide Michelson: • Narrow Michelson: 86 m. A (tuned) Tuning • Final width: 76 m. A Aperture Stop Blocking Filter Telescope lens set Telecentric Lens Michelson Narrowband Michelson ¼ Waveplate ½ Waveplates Lyot Waveplates ISS Beamsplitter and Limb Tracker Assembly BDS Beamsplitter Relay Lens Set Calibration lenses and Focus Blocks Front Window Filter CCD Shutter Assemblies CCD Fold Mirror SDO SUMMER SCHOOL, 2010 CCD Fold Mirror Page 9 of 587

Filter Profiles SDO SUMMER SCHOOL, 2010 Page 10 of 587

Framelist Example • Time: Time of first exposure at given wavelength since start of framelist execution • Tuning: I 1, I 2, … specify the tuning position • Doppler pol. : Polarization of image taken with Doppler camera • • – L and R indicate left and right circular polarization – Used for Doppler and line of sight field Vector pol. : Polarization of image taken with vector camera – 1, 2, 3, 4: Mixed polarizations needed to make vector magnetograms – Used for vector field reconstruction T data from the two cameras may be combined SDO SUMMER SCHOOL, 2010 Page 11 of 587

Observing Scheme • Observables made from filtergrams described by framelists • Filtergram properties – Wavelength – selected by rotating waveplates (polarizer for redundancy only) – Polarization state – Exposure time – Camera ID – Compression parameters, … – Determined by subsystem settings • • E. g. motor positions Framelists – Fixed list of filtergrams repeated at fixed cadence during normal operations – Entirely specified in software – highly flexible • But fixed scheme used throughout mission except for re-tunings and calibrations

Observables Calculation • Make I, Q, U, V, LCP, RCP from filtergrams – Identify missing and bad pixels – Correct for flat field and exposure time – Fill missing pixels – Correct for solar rotation and jitter (spatial interpolation) – Correct for acceleration effects (temporal interpolation) • Nyquist criterion almost fulfilled for Doppler and LOS • Nyquist is grossly violated for vector measurements in case of long framelists • Clever tricks exist – Apply demodulation matrix – Fill gaps, – Average in time, if desired • MDI-like and/or least squares for Doppler and LOS • Vector field inversion SDO SUMMER SCHOOL, 2010 Page 13

VFISV update Filter profiles VFISV HMI data VFISV description VFISV current status (Very Fast Inversion of the Stokes Vector) Forward modeling: Milne-Eddington approximation with Zeeman-induced Stokes profiles Filtering: Synthetic profiles are computed at high spectral resolution and then filtered with HMI’s filter profiles Inversion: Levengberg-Marquardt iterative scheme tries to match synthetic Stokes profiles to observed data in a leastsquares sense. Free parameters: , B, , , v. LOS, 0, D, , B 0, B 1 9/11/2021 Rebecca Centeno SDO SUMMER SCHOOL, 2010 Borrero et al 2009 14 SDO SWG meeting -- Sept 9, 2009

SDO SUMMER SCHOOL, 2010 Page 15

Summary of instrument properties • Filtergraph • 4096 x 4096 full disk coverage • 6173 Fe. I line • 0. 5” pixels, 1” optical resolution • 76 m. A filter profiles – Generally spaced at 69 m. A • Continuous coverage (>95%) • Doppler and LOS at 45 s cadence • Full Stokes at 45 s-135 s cadence – About 2 e-3 on (Q, U, V) in 135 s – About 1 e-3 in 12 minutes • Uniform quality • 95% temporal coverage – Eclipses are main problem SDO SUMMER SCHOOL, 2010 Page 16

Instrument Status – Current Observing (borrowed from J. Schou) • Doppler and LOS field at 45 s – One camera, 3. 75 s between images, 6 wavelengths and 2 polarizations – This will likely not change in the future • Vector field at 135 s – One camera, 3. 75 s between images, 6 wavelengths and 6 polarizations – Mostly binned to 12 minute averages – This may change in the future • • Eg. Go to 90 s cadence with only 4 polarizations Cross-calibration with MDI – Various MDI sequences – 60 day continues – 180 days low rate

HMI Observables • Continuum • Line depth • Line width • Dopplergram • LOS Magnetogram • Vector magnetic field SDO SUMMER SCHOOL, 2010 Page 20

Instrument Performance – Image Quality Spatial power spectra of snapshots of HMI Doppler data and MDI hires data integrated over azimuth. (courtesy: T. Duvall) SDO SUMMER SCHOOL, 2010 Page 21

Instrument Performance—comparison with MDI obs. 1. Left: Spectrum of 8 hours of HMI Doppler data near disk center; 2. Right: Spectrum of 8 hours of MDI hi-res data (from 1996); 3. Conclusion: ridges of HMI Doppler spectrum to high temporal and spatial frequencies. (courtesy: T. Duvall). SDO SUMMER SCHOOL, 2010 Page 22

Instrument Performance—Power Spectrum • Left: HMI Dopplergram power spectrum; • Right: HMI magnetogram power spectrum; • Conclusion: no significant leakage of p-mode to magnetograms (but yes for MDI magnetograms). Courtesy Tom Duvall SDO SUMMER SCHOOL, 2010 Page 23

Comparison-Polar Field (Line-of-sight Field on the Southern Polar Region) • Line-of-sight magnetic field on the southern polar region from HMI 720 -sec magnetogram (left) and MDI magnetogram (right). SDO SUMMER SCHOOL, 2010 Page 24

Magnetic Field Blos Zero Level • x 45 s SDO SUMMER SCHOOL, 2010 720 s Page 25

HMI vs MDI Blos Comparison 1. Data were taken roughly at the same time; 2. HMI data was degraded to MDI spatial resolution by convolving an 2 -D Gaussian function. SDO SUMMER SCHOOL, 2010 Page 26

Comparison—Line-of-sight Magnetograms SDO SUMMER SCHOOL, 2010 Page 27

MDI vs Solis MDI vs Mt. WSO HMI vs MDI HMI = 1. 03 SOLIS SDO SUMMER SCHOOL, 2010 Page 28

HMI Blos 45 vs HMI Blos Vector SDO SUMMER SCHOOL, 2010 HMI Blos 45 vs HMI Blos Vector for AR 1057 Page 29

Comparison—Magnetic Field Blos Mean Field HMI(45 s)—HMI(720 s)—MDI--WSO SDO SUMMER SCHOOL, 2010 Page 30

Instrument performance (summary) • Image quality – Optical quality is very good. Strehl ratio about 0. 85. – No significant field effects – Few bad pixels or other artifacts – Full well is excellent. Exposure levels in disk center continuum of 130 -150 ke- • Wavelength – Calibration looks very good – Drifts are slow (compared to MDI) – Artifacts are small • Polarization – Excellent polarimetric efficiency – Very low instrument polarization • Noise levels – LOS field: <10 G/pixel* – Doppler velocity about 20 m/s* • * We are not currently using all information. SDO SUMMER SCHOOL, 2010 Page 31

Conclusion: HMI is working extremely well. Total linear polarization SDO SUMMER SCHOOL, 2010 Page 32

HMI • Observables: – Full disk Doppler velocity (45 seconds cadence, 4096 x 4096 CCD); – Full disk light-of-sight magnetograms (45 seconds cadence, 4096 x 4096 CCD); – Full disk continuum intensity; and – Full disk vector magnetograms (90 seconds res. , 10 min cadence, 4096 x 4096 CCD). SDO SUMMER SCHOOL, 2010 Page 33

Framelist Choices Polarization Scheme • Options 1 and 2 – Cameras not combined – Same polarimetric noise per unit time – Option 1 (4 observations) is fast • Some Stokes parameters are made from differences over long time intervals (40 -50 s) • Significant acceleration effects – Option 2 (6 observations) is slower • • But all differences are close in time (4 s) • Almost no acceleration effects Options 3 and 4 – Cameras combined – Depends on ability to combine the cameras – Option 3 (remove duplicates from 1) only combines to make Vector • Better polarimetry than 1 and 2 • Calibrations on vector camera does not impact Doppler continuity – Option 4 (option one divided over cameras) combines to make both • • Even better polarimetry • Also better Doppler • But calibrations interrupt Doppler Time averaging helps SDO SUMMER SCHOOL, 2010 Page 34 of 587

On-Orbit Results – I -> (Q, U, V) leaks Mean -0. 070 e-3, slope: 15. 9 e-3 Mean -0. 199 e-3, slope: 8. 59 e-3 SDO SUMMER SCHOOL, 2010 Mean 0. 000 e-3, slope: -1. 34 e-3 Page 35 of 587

HMI Functional Specifications Summary SDO SUMMER SCHOOL, 2010 Page 36

HMI is not (yet) perfect We see the orbit in sensitivity, RMS, and Magnetic – we are trying to improve Slight degradation in throughput with time – we are watching Some periodic noise from tuning motor position noise – we are evaluating More – let us know… 9/11/2021 SDO SUMMER SCHOOL, 2010 37

Outstanding Issues • Image quality – • Wavelength Throughput drift • Contamination/radiation damage? – Refine MTF/OTF/PSF estimates • Possibly well enough to combine cameras – Focus drift – Half of information thrown away – Drifts slowly (compared to MDI) – Residuals vary with time of day – Unphysical variations – 7. 4 m. Hz artifact • Radiation damage? – • Adaptive algorithm? Improve flatfield • Polarization • Changes with time • Growing contamination • Bad pixels – Remove particle hits – Line formation issues Correlations between I and Q, U, V – MTF varies with state – Telescope polarization needs correcting – Ideal window temperature unknown • Affects depolarization • Radius and center moves – – Distortion needs refinement • Also waveplate induced – Scattered light needs to be measured – Corrupted images (3 in 2500000) – P-angle unknown – ISS optimization – Scattered light correction – Code needs refinement • Documentation – Three of four calibration papers submitted • But may need on-orbit refinement – Document data and processing Please help! Report artifacts! SDO SUMMER SCHOOL, 2010 Page 38