wide-field wavefront sensing in solar adaptive optics - its modeling and its effects on...

Wide-field wavefront sensing in Solar Adaptive Optics

- its modeling and its effects on reconstruction

Clémentine Béchet, Michel Tallon, Iciar Montilla, Maud Langlois

Centre de Recherche Astrophysique de Lyon (CRAL)Instituto de Astrofisica de Canarias (IAC)

May 2013, AO4ELT3 (Florence, Italy)

Outline

1. Context: AO for the 4-m diameter European Solar Telescope (EST)a) EST AO goalsb) preliminary design

2. Wavefront sensing in solar AOa) reminder on solar AO wavefront sensingb) anisoplanatism in wide-field sensor data

1. A model for wide-field wavefront sensinga) for more realistic data simulationb) for reconstruction?

2. Conclusions & Future work

Collaboration with Instituto Astrofisica de Canarias (I. Montilla)

Not a 30-m class telescope, just a 4-m one, but…– wide-field correction for high strehl at visible wavelengths– complexity close to TMT, GMT or E-ELT AO systems

construction 2019 -> 1st light 2025

EST AO goals : ambitious!– correction abilities over a wide range of elevations (down to

15 degrees)– requirements over a corrected fov of 1’ diameter:

Strehl (500nm) > 30% for r0 = 7cm Strehl (500nm) > 50% for r0 = 14cm Strehl (500nm) > 60% for r0 = 20cm

1. Context : AO for the 4-m diameter European Solar Telescope (EST)

EST AO system preliminary design– 5 DMs to cover a wide range of elevations– 1 on-axis (10”x10”) high-order Shack-Hartmann WFS– 1 “over-the-field” (70”x70”) low-order Shack-Hartmann WFS– sensed fields of 10”x10” for both SH-WFS– 1.5 -2 kHz frame-rate for cross-correlating sensed fields

designed from an budget error and Fourier code analysis (Berkefeld & Soltau, 2010)

… but requires end-to-end simulations

AO MCAO

DMs heights 0 km 0, 2, 6, 10, 23 km

DMs spacings 8 cm (~1800 act.) 8, 30, 30, 30, 30 cm (~ 4000 act.)

sensing fields 1 (10”x10”) 19 (10”x10”)

subap. size 8 cm (~1800 in total) 8 cm, 30cm (128 subaps. in total)

1. Context : AO for the 4-m diameter European Solar Telescope (EST)

Why end-to-end simulations?– lessons learned from E-ELT phase A studies, of huge difficulties

to provide an exhaustive and clear error budget, error term by error term, for tomography

– experienced limitation of Fourier codes

Octopus+FRiM-3D in closed-loop : E2EFRiM-3D reconstruction only: E2EOctopus+other reconstructor: E2E

Fourier code from ONERA

Le Louarn et al.SPIE 2012ATLAS/LTAO E2E simulations to determine the impact of LGS constellation radius on Strehl

Octopus : ESO E2E AO simulator

1. Context : AO for the 4-m diameter European Solar Telescope (EST)

Why end-to-end simulations?– lessons learned from E-ELT phase A studies, of huge difficulties to

provide an exhaustive and clear error budget, error term by error term, for tomography

– experienced limitation of Fourier codes

need of a fast simulator for very large AO systems– Octopus (ESO end-to-end simulator for AO)

cluster of hundreds of slaves diffractive model

– FRiM-3D (CRAL developments) sparse/fast modeling operators, geometric models, works even on an old laptop

started ESTAO E2E simulations (Montilla et al. SPIE 2012 and a poster this week) but needed new developments of FRiM-3D to adapt it to solar AO

=> focus of this talk: anisoplanatism in the AO wavefront sensing

1. Context : AO for the 4-m diameter European Solar Telescope (EST)

a) solar wavefront sensing in a nutshell– key approach = cross-

correlating SH-WFS– fov of 5”x5” minimum

(von der Lühe, 1983) to track gradients on granulation

– more robust if fov larger than 8’’x8’’

2. Wavefront sensing in solar AO

QuickTime™ et undécompresseur

sont requis pour visionner cette image.

(VTT example, von der Lühe et al., 2005)

a) solar wavefront sensing in a nutshell

– key approach = cross-correlating SH-WFS

– fov of 5”x5” minimum (von der Lühe, 1983) to track gradients on granulation

– more robust if fov larger than 8’’x8’’

combination of high-order WFS and 19 sensed fields (~10”x10”) in 70”x70” low-order WFS

but 10” is usually larger than the isoplanatic patch (500nm)

2. Wavefront sensing in solar AO

QuickTime™ et undécompresseur

sont requis pour visionner cette image.

(VTT example, von der Lühe et al., 2005)

b) Anisoplanatism effect on data

2. Wavefront sensing in solar AO

3”x3”50cm subap.

small & usually negligible fov in night-time AO

b) Anisoplanatism effect on data– cross-correlation over

10” fov = average gradient over the subap. and over the fov.

– independent contribution of each layers

– with increasing height, averaging over a wider layer area

– loosing sensitivity of the on-axis wavefront distortions at high altitude

2. Wavefront sensing in solar AO

10”x10”10cm subap.

wide-field solar AO sensing

b) Anisoplanatism effect on data– cross-correlation over 10”

fov = average gradient over the subap. and over the fov.

– independent contribution of each layers

– with increasing height, averaging over a wider layer area

– loosing sensitivity of the on-axis wavefront distortions at high height

– same conditions as Marino 2012 (Haleakala profile, no strong layer in high altitude) simulated with FRiM-3D

2. Wavefront sensing in solar AO

4m-diameter, 10cm subaps. reconstruction error only

(noiseless)

fitting

b) Anisoplanatism effect on data– cross-correlation over 10’’

fov = average gradient over the subap. and over the fov.

– independent contribution of each layers

– with increasing height, averaging over a wider layer area

– loosing sensitivity of the on-axis wavefront distortions at high height

– same conditions as Marino 2012 (Haleakala profile, no strong layer in high altitude) simulated with FRiM-3D

2. Wavefront sensing in solar AO

4m-diameter, 10cm subaps. reconstruction error only

(noiseless)

fitting

We need to model the wide-field wavefront sensing in solar E2E simulations

a) for more realistic simulations by FRiM-3D– “small fov” model already from night-time AO– based on continuous interpolation functions to model the

turbulent layers

– “wide-field” solar model : includes the average over the fov F, introducing enough subdirections for sensing

3. A model for wide-field wavefront sensing

contribution per layer

ideally 0”

new!

a) for more realistic simulations by FRiM-3D anisoplanatism contribution to data, per layer :

3. A model for wide-field wavefront sensing

a) for more realistic simulations by FRiM-3D anisoplanatism contribution to data, per layer :

for a complete profile, contribution of layers can be added in variance

for examples:– E-ELT profile (MAORY phase A)– Cerror Pachon median profile– Haleakala profile (Marino 2012)

– wfs fov = 8” x 8”– r0 = 15cm– zenith angle = 60 deg.

~ 40nm rms

3. A model for wide-field wavefront sensing

b) for the reconstruction? FRiM-3D can use various fast models of SH sensing in its

minimum-variance reconstruction algorithm

3. A model for wide-field wavefront sensing

b) for the reconstruction? FRiM-3D can use various fast models of SH sensing in its

minimum-variance reconstruction algorithm

“small-fov” “wide-fov”

with nsubdir = 4

10”

10”

10”

10”only 1 reconstructed layer

several reconstructed layers

3. A model for wide-field wavefront sensing

b) for the reconstruction? FRiM-3D can use various fast models of SH sensing in its

minimum-variance reconstruction algorithm

“small-fov” “wide-fov”

with nsubdir = 4

10”

10”

10”

10”only 1 reconstructed layer

several reconstructed layers

3. A model for wide-field wavefront sensing

is there a benefit in using S4 in the reconstruction?

b) for the reconstruction?§ toy case #1 : 1 simulated layer only

3. A model for wide-field wavefront sensing

fitting

b) for the reconstruction?§ toy case #1 : 1 simulated layer only

3. A model for wide-field wavefront sensing

fitting fitting

b) for the reconstruction?§ toy case #1 : 1 simulated layer only

3. A model for wide-field wavefront sensing

probable benefit since S4 allows to reconstruct layers in altitude…

fitting fitting

b) for the reconstruction?§ toy case #2 : 1 simulated layer only2 reconstructed layers: at 0 km and another height h

3. A model for wide-field wavefront sensing

fitting

b) for the reconstruction?§ toy case #2 : 1 simulated layer only2 reconstructed layers: at 0 km and another height h

3. A model for wide-field wavefront sensing

???

fitting

b) for the reconstruction?§ toy case #2 : 1 simulated layer only2 reconstructed layers: at 0 km and another height h

3. A model for wide-field wavefront sensing

???

benefit from S4 is strongly related to the priors!

fitting fitting

b) for the reconstruction?§ Complete atmosphere reconstruction (Haleakala profile,

Marino 2012)

3. A model for wide-field wavefront sensing

at the end of the day, no quantitative benefit from S4 in single-sensor reconstruction

checked for various profiles, elevations, seeing, noise levels, …

the average of the data over the field of view definitely degrade the possible performance

4. Conclusions & future work challenging AO system for the future European solar telescope

simulations of the system started in collaboration between CRAL and IAC

to understand new issues of solar AO (e.g. anisoplanatism for SCAO) to consolidate the error budget (anisoplanatism, 5 DMs, low-order

sensing, high-order sensing)

simulator FRiM-3D, developed at CRAL, now allows to model the anisoplanatism in the wide-field data

anisoplanatism error can range from a few nm rms to 100 nm rms (for bad conditions)

the use of an approximated model with 4 subdirections in the reconstructor does not significantly reduces the drop of the SCAO Strehl due to the sensing anisoplanatism

anisoplanatism can now however be simulated in solar MCAO mode (on-going EST and Big Bear simulations with FRiM-3D), and we will evaluate how much it could be canceled by the tomographic correctionPosters: I. Montilla et al. (ESTAO), M. Langlois et al. (Big Bear)