Interpreting HMI multi-height velocity measurements Kaori Nagashima Postdoc of Interior of the Sun and Stars Dept. @MPS (May 2012 - ) Collaborators of this study: L. Gizon, A. Birch, B. Löptien, S. Danilovic, R. Cameron (MPS), S. Couvidat (Stanford Univ.), B. Fleck (ESA/NASA), R. Stein (Michigan State Univ.) 2013.11.19. Solar Group Seminar @MPS 1
Interpreting HMI multi-height velocity measurements Motivation Multi-height velocity info is useful in many purposes: Study of energy transport in the solar atmosphere (e.g., Jefferies et al. 2006, Straus et al. 2006) Detection of flows in the chromosphere using multi-line observations by helioseismology technique (e.g., Nagashima et al. 2009, see next slides) If we can obtain multi-height velocity info from HMI full-disc every-day observations, it has advantage in that we have much larger amount of datasets available compared with any other current observations. Want to obtain multi-height velocity info from SDO/HMI observation datasets! 2
Multi-wavelength helioseismology study example: Helioseismic signature of chromospheric downflow in acoustic travel-time measurements from Hinode Measure acoustic travel time in AR and in QS [Mm] Use photospheric (ph) and chrospheric (ch) datasets In QS supergranular patterns are seen both in ph and ch. In AR, only in chromospheric datasets travel time anomaly is detected (Nagashima et al. 2009 ApJL) Outward travel time<inward travel time sample images outward-inward travel-time difference maps [Mm] Ca II H ch black : outward <inwar white : outward>inwar grayscale:-1 ~+1 min Fe Doppler ph [Mm] [Mm] They are different!!! 3
What we could say by the multi-height helioseismology was In an emerging flux region (EFR), we found travel time anomaly in plage in chromosphere is stronger than in photosphere. This can be interpreted as DOWNFLOWS in chromosphere. Emerging flux Downflow V~2km/s V <1km/s? sunspots plage chromosphere photosphere magnetic field line 4
Interpreting HMI multi-height velocity measurements Motivation Multi-height velocity info is useful in many purposes: Study of energy transport in the solar atmosphere (e.g., Jefferies et al. 2006, Straus et al. 2006) Detection of flows in the chromosphere using multi-line observations by helioseismology technique (e.g., Nagashima et al. 2009, see next slides) If we can obtain multi-height velocity info from HMI full-disc every-day observations, it has advantage in that we have much larger amount of datasets available compared with any other current observations. Want to obtain multi-height velocity info from SDO/HMI observation datasets! 5
some attempts to obtain multiheight info from SDO/ HMI Fleck et al. (presentations @AGU 2010 etc.) Report the phase difference in their multi-height Dopplergrams made by HMI filtergrams Rajaguru et al. (2012) Exploit the multi-height HMI and AIA data to study power enhancement around ARs in various heights. So. It is promising. Fleck et al. (a figure in their poster at AGU in 2010) downward propagating phase 6 Atmospheric gravity mode signature
Create multi-height Dopplergrams using SDO/HMI observables 7
Helioseismic and Magnetic Imager (HMI) onboard Solar Dynamics Observatory (SDO) HMI observes the Sun in Fe I line at 6173 Å HMI takes filtergrams at 6 wavelengths around the line. SDO HMI Standard Dopplergram is derived from these 6-wavelength filtergrams basically the center of gravity of the line (see next slide) I5 at-172.0ma I4 at -103.2mA I3 at -34.4mA I5 I0 at +172.0mA I1 at +103.2mA I2 at +34.4mA I0 Fe I line profile In this work, using these filtergrams, we try to make multi-height Dopplergrams instead. λ HMI filter tuningposition profiles Fig. 6 in Schou et al. 2011 SoPh 8
Standard HMI Dopplergram (Couvidat et al. 2012) Calculate the line shift based on the Fourier coefficients of the 6 filtergrams I5 I0 λ Considering the line asymmetry etc., they calibrate this v by using calibration table, and make the standard Dopplergrams (pipeline products) Formation layer @ ~100km above the surface (Fleck et al. 2011) Similar to the formation layer of the center of gravity of the 6 filtergrams. 9
At first, we made 3 simple Dopplergrams, but it did not work well. Doppler signal: I b I r V = f I b I r I b +I r I b +I r core, wing, far-wing Deeper layer Far-wing Wing Core fitting the average Doppler signals by 3rd order polynomial using the SDO orbital motion Disadvantage 1:SDO motion (and fitting range) is limited (<3.5km/s) Disadvantage 2: wavelength separation (and dynamic range) is limited If the velocity exceeds 1.7km/s, the line center is outside of the core pair, and cannot use this method. Details: Nagashima et al. 2013 (accepted for ASP conference series) I b I r Shallower layer 10
SDO velocity [m/s] core I 3 saturated Usable only within a limited range I 2 Doppler signal averaged over FOV I 3 I 2 I 3 + I 2 If v = 1. 7km/s Δλ = 33. 4mA Limited valid range due to small wavelength separation 11
We tried several other definitions of Dopplergrams, and found these two look good. 1. Average wing (for deeper layer) Calculate the Doppler signals using the average of each blue and red wing. I b I r ( I I b +I b = I 5+I 4 r 2, I r = I 0+I 1 2 ) I5 I4 I0 I1 λ Convert the signal into the velocity: 1. Calculate the average line profile 2. Parallel-Dopplershift the average line profile 3. Calculate the Doppler signals 4. Fit to a polynomial function of the signal λ 12
We tried several other definitions of Dopplergrams, and found these two look good. 2. Line center (for shallower layer) Doppler velocity of the line center derived from 3 points around the minimum intensity wavelength Calculate the parabola through the 3 points and use its apex as the line shift So, we have 1. Average-wing Dopplergrams 2. Line-center Dopplergrams 3. And Standard HMI Dopplergrams (pipeline products) Now we have 3 Dopplergrams! λ Are they really multi-height? 13
Are they really multi-height Dopplergrams? (1) Estimate of the formation height using simulation datasets 14
Are they really multi-height Dopplergrams? (1) Estimate of the formation height using simulation datasets 1. Use the realistic convection simulation datasets: STAGGER (e.g., Stein 2012) and MURaM (Vögler et al. 2005) 2. Synthesize the Fe I 6173A absorption line profile using SPINOR code (Frutiger et al. 2000) 3. Synthesize the HMI filtergrams using the line profiles, HMI filter profiles, and HMI PSF 4. Calculate these Dopplergrams: Line center & Average wing & standard HMI 5. Calculate correlation coefficients between the synthetic Doppler velocities and the velocity in the simulation box 15
Sample filtergram images (10Mm square) HMI observation data ~370km/pix STAGGER synthetic filtergrams (reduced resolution using HMI PSF, ~370km/pix) STAGGER synthetic filtergrams (with STAGGER original resolution, 48km/pix) 16
Sample synthetic Dopplergrams (10Mm square) HMI observation STAGGER synthetic filtergrams (reduced resolution using HMI PSF, 3.7e2km/pix) Standard HMI Dopplergram STAGGER synthetic filtergrams (with STAGGER original resolution, 48km/pix) Average wing Line center Synthetic HMI Dopplergram 17
Estimate of the formation height using simulation datasets Correlation coefficients between the synthetic Doppler velocities and the velocity in the simulation box Correlation coefficients Peak heights: Line center 221km Standard HMI 195km Average wing 170km Line center Standard HMI Average wing 26km 25km 18
Estimate of the formation height using simulation datasets Correlation coefficients between the synthetic Doppler velocities and Correlation coefficients the velocity in the simulation box (with original STAGGER resolution (no HMI PSF)) w/ PSF they are higher! Peak heights: Line center 144km Standard HMI 118km Average wing 92km Line center Standard HMI Average wing 26km 25km 19
Estimate of the formation height using simulation datasets Correlation coefficients between the synthetic Doppler velocities and the velocity in the simulation box Correlation coefficients MURaM simulation data 17.6km/pix Peak heights: Line center 150km Standard HMI 110km Average wing 80km Line center Standard HMI Average wing 40km 30km 20
Vz auto-correlation coefficient in the wavefield provided by STAGGER datasets The width of the correlation peak is so large. STAGGER (original resolution) STAGGER (w/ HMI PSF) Therefore, the Dopplergram of this wavefield should have such a wide range of contribution heights. Wide peaks 21
Contribution layer is higher when the resolution is low (i.e.,w/ PSF) If the formation height in the cell is higher In the cell it is brighter than on the intergranular lane The cell contribution is larger than the intergranular lane s contribution? Therefore, the contribution layer is higher. right??? STAGGER simulation data a) Continuum intensity map b) Surface vertical velocity map c) τ = 1 layer height map 22
Are they really multi-height Dopplergrams? (2) Phase difference measurements 23
Power maps of the Dopplergrams HMI observation data STAGGER simulation data Line center Standard HMI Dopplergrams Averagewing Horizontal wavenumber x Rsun *No data due to the different cadence (1-min for STAGGER, 45-sec for HMI obs) 24
Phase difference between Doppler velocity datasets from two different height origins No significant phase difference (in p- mode regime) Line center HMI Average wing HMI observation data Atmospheric gravity wave? (e.g., Straus et al. 2008, 2009) a b The waves above the photospheric acoustic cutoff (~5.4mHz) can propagates upward. -> Phase difference between two layers with separation Δz Δφ 2ππ = Δz c s Significant phase difference is seen. Surely they are from different height origin. Rough estimate: Photospheric sound speed: c s ~7 km/s Phase difference measured: Δφ = 30 deg @8mHz Δz~ 73km This means what? 25
We have estimated the contribution layers by calculating the correlation coefficients between the Doppler velocities and Vz in the atmosphere That was for bulk velocities. Here by the phase difference map in the k-ω space, we consider each (k, ω) component. In this case, the velocity for each component is small (can be considered as linear perturbation from the total velocity) Here we try to use response function 26
Response function convolved with the HMI filter profiles Def: Calculated by STPRO in SPINOR code (Frutiger et al. 2000) I λ, v z I λ, v z Iconst response function = ddd z, λ vv z z v z z I λ, v z : Intensity at the wavelength λ if the velocity field is v z = v z (z) z: geometrical height (z = 0 @ τ 5000A = 1) I5 I0 Height [km] 27
Response functions for For simplicity, here we consider only for the simple Dopplergrams,D br = I b I r I b +I r And assume response function for D bb is R~R λ b R λ r. Difference between averagewing and core (substitute for line center ) is 44km simple Dopplergrams Center-ofgravity heights 147.4km 166.6km 143.9km The height difference roughly estimated by the phase difference is Δz~ 73km 191.8km 28
Phase difference between Doppler velocity datasets from two different height origins Line center HMI Average wing HMI observation data a b *No data due to 1-min cadence (STAGGER, 45-sec for HMI obs) STAGGER simulation data In the STAGGER simulation data Acoustic cutoff frequency seems lower (<5.4mHz) No signals in the atmospheric gravity wave (still not sure why, but maybe 29 due to lack of temperature minimum? Or lack of thick atmosphere?)
Phase difference of Vz in different height layers STAGGER Vz STAGGER synthetic Dopplergrams Similar to the synthetic Dopplergrams 170km 144km 118km 92km a b c Line center HMI Average wing a 30 b
Phase difference (CO 5 BOLD case) COBOLD IBIS obs. Phase difference of the velocity fields at 250km and 70km above surface They have -negative phase shift above the acoustic cutoff Fig. 1 in Straus et al. 2008 - Positive phase shift in the lower frequency ranges (atmospheric gravity waves) 31
Summary Confirm that we can obtain multi-height velocity information in the solar atmosphere using SDO/HMI data By estimating the contribution layer height of the multi-height velocity info using STAGGER/MURaM simulation datasets Line center Standard HMI 30km 30-40km Average wing By calculating the phase difference between the velocities with different height origins. Note: We limit these discussions in the Quiet Sun. 32