Solar Ground-Layer Adaptive Optics

Transcription

1 PUBLICATIONS OF THE ASTRONOMICAL SOCIETY OF THE PACIFIC, 127: , 2015 May The Astronomical Society of the Pacific. All rights reserved. Printed in U.S.A. Solar Ground-Layer Adaptive Optics DEQING REN, 1 LAURENT JOLISSAINT, 2 XI ZHANG, 3 JIANPEI DOU, 3 RUI CHEN, 3 GANG ZHAO, 3 AND YONGTIAN ZHU 3 Received 2014 November 01; accepted 2015 March 20; published 2015 April 6 ABSTRACT. Solar conventional adaptive optics (CAO) with one deformable-mirror uses a small field-of-view (FOV) for wave-front sensing, which yields a small corrected FOV for high-resolution imaging. Solar activities occur in a two-dimensional extended FOV and studies of solar magnetic fields need high-resolution imaging over a FOV at least 60. Recently, solar Tomography Adaptive Optics (TAO) and Multi-Conjugate Adaptive Optics (MCAO) were being developed to overcome this problem of small AO corrected FOV. However, for both TAO and MCAO, wavefront distortions need to be tomographically reconstructed from measurements on multiple guide stars, which is a complicated and time-consuming process. Solar Ground-Layer Adaptive Optics (S-GLAO) uses one or several guide stars, and does not rely on a tomographic reconstruction of the atmospheric turbulence. In this publication, we present two unique wavefront sensing approaches for the S-GLAO. We show that our S-GLAO can deliver good to excellent performance at variable seeing conditions in the Near Infrared (NIR) J and H bands, and is much simpler to implement. We discuss details of our S-GLAO associated wavefront approaches, which make our S-GLAO a unique solution for sunspot high-resolution imaging that other current adaptive optics systems, including the solar MCAO, cannot offer. Online material: color figures 1. INTRODUCTION The study of two-dimensional solar fine structures requires high-resolution imaging over a large FOV. A sunspot may have an angular extension up to 60. One of the current and future developments of solar adaptive optics is MCAO. In a MCAO system, several deformable mirrors (DMs), each conjugated to a different height above the telescope aperture, are used so that the atmospheric turbulence at different altitudes can be corrected differentially and high-resolution imaging is available over a large FOV (Johnston & Welsh 1994; Ragazzoni et al. 1999; Tokovinin & Viard 2001; Ellerbroek et al. 2003). Although MCAO is a promising technique for high-resolution imaging over a large FOV, it is a challenging technique, since this technique requires tomographic wavefront reconstruction, which is a highly complicated and time-consuming process. To accurately reconstruct the wavefront or turbulence profile at different altitudes, a number of guide stars are needed. Two different MCAO techniques exist; both deploy multiple DMs conjugated to different heights. One is the star-oriented MCAO, in which a number of guide-stars (for solar wavefront sensing, a region with a FOVas small as 8 8 can be used as a guide star) are used for wave-front sensing and the wavefront 1 Physics and Astronomy Department, California State University Northridge, Nordhoff Street, Northridge, CA 91330; ren.deqing@csun.edu. 2 Leiden Observatory, Niels Bohrweg 2, NL-2333 CA Leiden, The Netherlands. 3 National Astronomical Observatories/Nanjing Institute of Astronomical Optics and Technology, Chinese Academy of Sciences, Nanjing , China. profile can be reconstructed via tomography (Ragazzoni et al. 1999). The resulting reconstructed wavefront information is used to control the DMs; the other is the layer-oriented approach (Ragazzoni et al. 2000). The layer-oriented MCAO was initially introduced for nighttime observations, in which each wave-front sensor and its associated DM are conjugated to a specific height above the telescope, and no tomographic wave-front reconstruction is needed, which greatly simplifies the system; Kellerer (2012) proposed the use of layer-oriented MCAO for solar high-resolution imaging. Marino and Wöger (2014) pointed out that in solar layer-oriented MCAO, the Shack-Hartmann wavefront sensor (SHWFS) images are vignetted. As explained in Kellerer (2014), this requires an adjustment of the correlation algorithms to determine the local wavefront slopes. Marino and Wöger further suggest that the method is not viable due to field reduction at high-conjugate layers. This issue is likewise addressed by Kellerer (2014): Field reduction is inherent to the layer-oriented approach. It particularly affects nighttime systems because the signal is averaged over a limited number of stars, rather than a continuous field. Since night-time systems are successfully used on-sky, the field reduction will be even less critical in the solar application. The S-GLAO can be viewed as a layer-oriented MCAO system that deploys only one DM conjugated on the telescope aperture; therefore, there is no field vignette or field reduction. Rimmele et al. (2010) tested an S-GLAO system with the National Solar Observatory (NSO) Dunn Solar Telescope (DST): they used a wavefront sensor conjugated to the ground and averaged the wavefront distortions 469

2 470 REN ET AL. FIG. 1. Two different wave-front sensing approaches for the S-GLAO: Left: discrete guide regions, and right: a large continuous guide star region. over a FOV, and the wavefront data were used to control the deformable mirror. The experiment, however, was unsuccessful. Due to the complexity of MCAO, there is no routine operational solar MCAO available until now, although some progress has been made recently. The MCAO systems that are currently being developed for the 1.6-m NST and the 1.5-m GREGOR or the future 4-m Daniel K Inouye Solar Telescope (DKIST) and European Solar Telescope (EST) target for a corrected imaging FOV in the visible wavelength range (Berkefeld & Soltau 2010; Rimmele & Marino 2011; Schmidt et al. 2014). Theoretical simulation of the future EST MCAO system using 5 DMs and 19 guide stars at seeing r 0 ¼ 15 cm for zenith angle = 0 shows that Strehl ratios on the range of 0:2 0:59 can be achieved in a FOV of 60 diameter, at the 0:5-μm visible wavelength (Berkefeld & Soltau 2010). Recently, solar TAO, which uses one deformable mirror and multiple guide stars, was proposed to overcome the problem of the small CAO corrected FOV (Ren et al. 2014a). The TAO is optimized for the near infrared (NIR) J and H band high-resolution imaging over a FOV of 60, where the NIR spectrums are more sensitive for the solar magnetic field measurements than that in the visible (Ramsauer et al. 1995; Lin & Rimmele 1999; Khomenko et al. 2003), and where the TAO will be able to deliver comparable Strehl ratios in the NIR with that of the MCAO at the 0:5-μm visible. However, for the TAO, the wavefronts still need to be reconstructed tomographically from measurements based on multiple guide stars, which is a complex and time-consuming FIG. 2. The asterism of 4 guide stars for DGS wavefront sensing. FIG. 3. SHWFS vignetting as a function of conjugated height.

3 SOLAR GROUND-LAYER ADAPTIVE OPTICS 471 TABLE 1 DISCRETE SEEING PARAMETER r 0 ðhþ PROFILE AT DIFFERENT SEEING CONDITIONS Height (m) r o (m) at average seeing r o (m) at bad seeing r o (m) at good seeing process. In addition, because of the large size (up to several tens of arcseconds) and high-contrast of the sunspot image, the multiple-guide-star wavefront sensing approach is difficult or impossible to use a sunspot for wavefront sensing (since the sunspot large dark area cannot be used as guide stars). Current solar MCAO systems are all based on multiple-guide-star wavefront sensing, which makes them difficult to use the highcontrast sunspot images for wave-front sensing. In the case of sunspot high-resolution imaging, these MCAO systems must use the nearby solar granulation for wave-front sensing, which suffers from the problem of so called isoplanatic angle and thus sunspot imaging performance will be degraded rapidly because of the off-axis wavefront sensing angle. In this publication, we present our S-GLAO wavefront sensing approaches, which does not require a tomographic wavefront reconstruction. In addition to using several discrete guide stars for wave-front sensing, our wavefront approaches can also use the entire sunspot for wavefront sensing, which provides a unique solution for sunspot high-resolution imaging that other systems cannot offer. Our past experiences (Ren & Dong 2012; Ren & Zhu 2013; Ren et al. 2014b) with sunspot wavefront sensing technique indicates that such a S-GLAO system with the proposed wavefront sensing approaches will be able to deliver extremely stable wavefront corrections at different seeing conditions, including the extreme bad seeing conditions with the Kitt Peak NSO 1.6-m McMP telescope where the typical daytime seeing parameter r 0 is only 5 cm (Keller 2005; Ren and Zhu 2013), compared with other sites where the average seeing parameter r 0 is 7 cm (Socas-Navarro et al. 2005). To our knowledge, our S-GLAO is the only adaptive optics system that can still provide effective wave-front correction for sunspot high-resolution imaging in such bad seeing conditions, which will significantly improve AO observation efficiency. While the S-GLAO can deliver very promising performances, it is simple to implement. We discuss details of our unique wavefront sensing approaches and provide associated S-GLAO estimated performances. 2. S-GLAO SYSTEM 2.1. S-GLAO Wavefront Sensing Approaches Solar adaptive optics uses a SHWFS to measure the local slope in each subaperture. For our S-GLAO, two different wavefront sensing approaches are available: one is the use of several discrete guide stars (called DGS herein), and the other is the use of a large FOV such as as a large guide star (called LGS herein). For the DGS approach, a small FOV region on the order of 8, smaller than the isoplanatic angle is used as a guide star in the NIR, and all the wavefronts measured by a SHWFS from these guide stars are averaged, and the averaged wavefront is corrected by a deformable mirror conjugated at the same height with the SHWFS. This is equivalent to the average of the slopes in each subaperture from these guide stars. To achieve additional performance, the S-GLAO SHWFS and DM can be conjugated to an appropriate height. For LGS approach, it is exactly identical to conventional solar AO, except that the guide star is corresponding to a large FOV region in the order of For this approach, the wavefront distortions are averaged optically over the large FOV via the cross-correlation of SHWFS images. In principle, the DGS is equivalent to the LGS, provided a large number of guide stars are used for wavefront sensing (enough guide star regions to fill the entire field). The DGS is suitable to be applied on a uniform FOV such as solar granules, as shown in Figure 1 (left), in which 4 guide stars (represented by four small squares, each with a region of 8 8 FOV) are used and can be placed anywhere in the wavefront sensing FOV. The LGS has the advantage of being applicable on sunspots, as shown in Figure 1 (right), in which a large guide region (represented by the large square region with a FOV) enclosed the entire sunspot, is used as a guide star for wave-front sensing. Of course, the LGS approach can also use solar granulation as a guide region. Since a sunspot has a typical size of several tens of arcseconds and because it is a high-contrast image, DGS approach cannot be employed for sunspot wavefront sensing, at least under average or poor seeing condition. TABLE 2 STREHL RATIOS AT DIFFERENT OFOVS AT1:25-μm WAVELENGTH AND AVERAGE SEEING CONDITION (r 0 ¼ 70 mm), WITH 0CONJUGATE HEIGHT SR: CAO SR: 40 OFOV SR: 60 OFOV Note. Both the imaging FOV and OFOV are defined in diameter.

4 472 REN ET AL. TABLE 3 STREHL RATIOS WITH DIFFERENT OFOVS AT1:25-μm WAVELENGTH AND AVERAGE SEEING CONDITION (r 0 ¼ 70 mm), WITH 400-M CONJUGATION HEIGHT SR: CAO SR: 40 OFOV SR: 60 OFOV To our knowledge, solar AO correction with a single DM and wave-front sensing based on the LGS has never been used, although multiple guide stars are being applied for solar MCAO systems using star-oriented approach to reconstruct the tomographic wavefront. In our DGS approach, we use discrete multiple guide stars, from which the average slope are calculated and are fed directly to the conventional AO software, which avoids the tomographic wavefront reconstruction that is time consumed. Solar AO correction with a wavefront sensing FOV was tested on the 0.7-m Dunn solar telescope by Rimmele et al. (2010). However, it was unsuccessful, because of the poor SHWFS sampling accuracy (at 2 per SHWFS detector pixel), which cannot provide acceptable AO performance. However, our recent AO systems with subpixel wavefront sensing accuracy successfully demonstrated that such a system can deliver good performances over a large imaging FOV, and such an AO system is able to deliver an extremely stable wavefront correction at different seeing conditions (Ren & Dong 2012; Ren & Zhu 2013; Ren et al. 2014b). Of course, because of the large wavefront sensing FOV, fast computation or image processing speed is required. However, the AO running speed is a less serious issue in the NIR than in the visible. To our knowledge, our S-GLAO is the only system that is able to use the entire sunspot as a guide star region and is able to deliver an extremely stable atmosphere turbulence correction at the Kitt Peak bad seeing conditions (with an average Fried parameter r 0 ¼ 5 cm), which clearly demonstrates that our S-GLAO with the LGS wavefront sensing approach is an optimized solution for sunspot high-resolution imaging. For our DGS wavefront sensing, we will use 4 guide stars for the S-GLAO performance simulations, with the layout shown in Figure 2. The 4 guide stars are uniformly placed in the wavefront sensing FOV. Since an increase in a few number of guide stars will not significantly improve the AO performance, the 4 guide stars are viewed as a good trade-off between the AO running speed and performance. For the LGD wavefront sensing, for the S-GLAO performance evaluations, we will use a large guide region with a FOV of 40 or 60 in diameter, respectively, which is achievable with our current AO technique in view of computation power requirement and wavefront sensor camera running speed. As discussed by Marino and Wöger (2014) and Kellerer (2014), when a SHWFS is conjugated to a distance away from the telescope aperture, the SHWFS images are vignetted and the effective wavefront sensing FOV is reduced. Here, we present a quantitative calculation for the vignetting and FOV reduction, and this process is shown in Figure 3. Assume that d is the subaperture diameter, which is chosen according to the seeing condition. Increasing d means that only lower order wavefront distortions can be sensed, and this reduces the AO performance. We assume that d is fixed in our wave-front sensing process. Guide stars 1 (indicated by star 1) and 2 (indicated by star 2) are located on on-axis and the boundary of the wavefront sensing FOV, respectively. Guide star 2 corresponds to the maximum off-axis angular radius of θ used for the wavefront sensing. For clarity, only one subaperture is shown. When located on the telescope aperture (i.e., on the ground with zero conjugated height), all the light within the SHWFS FOV passes through the subaperture, and no vignetting is induced. For our S-GLAO wavefront sensing, the SHWFS and the corresponding DM can be conjugated to an appropriate conjugate height, where most ground layer turbulence is concentrated (for the daytime solar observation, it is close to the ground), and this can provide a better wavefront correction. For better wavefront sensing, we Strehl ratio FOV (arc second) FIG.4. Strehl ratios of DGS-based system with different OFOVs at 1:25 μm NIR at average seeing conditions. Solid line represents the CAO; dotted line, optimized for 40 OFOV; dash-dotted line, optimized for 60 OFOV. See the electronic edition of the PASP for a color version of this figure.

5 SOLAR GROUND-LAYER ADAPTIVE OPTICS 473 TABLE 4 STREHL RATIOS AT DIFFERENT OFOVS AT1:25-μmWAVELENGTH AND GOOD SEEING CONDITIONS (r 0 ¼ 91 mm), WITH A 400-M CONJUGATE HEIGHT SR: CAO SR: 40 OFOV SR: 60 OFOV simply define 50% vignetting of the marginal ray as the threshold at which the cross-correlation can still be conducted effectively. That is, wavefront sensing can be performed only when the vignetting of marginal rays in the SHWFS FOV is less than or equal to 50%. Of course, for other points in the wavefront sensing FOV, vignetting will be smaller. Although this wavefront sensing criterion may be too strict, it will guaranty that the vignetting has a negligible effect for our S-GLAO wavefront sensing. From Figure 3, we can immediately derive the allowed maximum conjugated height h, which corresponds to a 50% vignetting. The maximum conjugated height h is a function of the wavefront sensing FOV θ, subaperture diameter d, with h ¼ðd=2Þ tanð90 θþ. One can continue to move the SHWFS upward to a conjugated height H, where the marginal rays are exactly totally vignetted or disappeared (i.e., wavefront FOV reduction occurs) in the subaperture. From Figure 3, it is clear that the conjugated H at which this occurs is H ¼ d tanð90 θþ. At conjugated H, the corresponding guide star 2 has no any contribution for wavefront sensing, and thus cannot be used for wavefront sensing. Therefore, a guide star corresponding to a conjugated height equal or larger than H cannot be used for solar wavefront sensing. In fact, at a conjugated height in the range between h H, the corresponding guide region may still be used for wavefront sensing, at the cost of tolerating the corresponding vignetting. The above discussion regarding the vignetting or field reduction is applicable to both DSG and LGS wavefront sensing approaches Performance Simulations For S-GLAO performance simulations, we use the Big Bear Solar Observatory (BBSO) site seeing measurement data, which were presented in Kellerer et al. (2012) recently. The BBSO experience can be summarized as follows: most of the turbulence is concentrated in four layers, the ground layer (0 500 m), the extended ground layer (1 2 km), the boundary layer (3 7 km), and the tropopause ( 8 km). In our previous publication (2014a), we interpolate the four-layer measurement data as a 10-layer profile, which may result in conservative results for performance simulations. In this publication for our one-dm S-GLAO system, we directly use the four-layer measured seeing profile data, which should give more realistic simulation results. The seeing conditions at BBSO are, generally, good in summer and bad in winter. Table 1 shows the distribution of the Fried parameter r 0 ðhþ at different heights h above the telescope aperture. At different seeing conditions (good, bad, and average), we assume that the four turbulent layers are concentrated at heights 200, 1000, 3000, and 8000 m. The Fried parameter r 0 ðhþ is slightly adjusted to be consistent with the isoplanatic angle θ 0, which is 6,2, and 4 at good, bad, and average seeing conditions, respectively. From this table, it is clear that for the daytime solar observation, most turbulence (i.e., ground and extended grounds) is concentrated at a height very close to the ground. Therefore, an effective correction of this turbulence at an appropriate conjugated height with a DM should be able to provide a good performance over a large imaging FOV in the NIR J and H long wavelengths, at least at average and good seeing conditions. In our previous work with TAO (Ren et al. 2014a), we found that a one-dm AO system will not be able to provide acceptable performance over a large imaging FOV in the visible. However, it does offer good performance in the NIR J and H bands. From the scientific point of view, solar activities are dominated by magnetic fields and NIR spectral lines are more sensitive for solar magnetic field measurements than at the visible light. While work using visible spectral lines tends to find the kg fields (Grossmann-Doerth et al. 1996; Socas-Navarro & Sánchez Almeida 2002), observational studies based on the TABLE 5 STREHL RATIOS AT DIFFERENT OFOVS AT1:65-μm WAVELENGTH AND AVERAGE SEEING CONDITIONS (r 0 ¼ 70 mm), WITH A 400-M CONJUGATE HEIGHT. SR: CAO SR: 40 OFOV SR: 60 OFOV

6 474 REN ET AL. Strehl ratio FOV (arc second) FIG. 5. The same as Fig. 4, but at 1:65 μm NIR wavelength. See the electronic edition of the PASP for a color version of this figure. infrared spectral lines find the magnetic field strength in the range below 1 kg (Lin & Rimmele 1999; Khomenko et al. 2003). Since solar high-resolution imaging is mostly done at average and good seeing conditions, in the remaining of this article, we will only calculate the S-GLAO performances at the NIR J and H bands at average and good seeing conditions. The software we used for our S-GLAO performance simulations is PAOLA an Astronomical Adaptive Optics Modeling Toolbox, written by Laurent Jolissaint (Jolissaint 2010). PAOLA is one of five well-recognized software packages which are dedicated for adaptive optics performance simulations, and these codes were thoroughly tested and compared to one another to ensure a high degree of confidence in the results (Andersen et al. 2006). PAOLA is free to download and is used by many research groups for AO performance simulations over the world. The DGS-based S-GLAO performances are calculated by using PAOLA s GLAO star modes, in which the phase is known in the direction of a discrete number of NGS, while the LGSbased S-GLAO performances are evaluated by using PAOLA s GLAO full modes in which the phase is assumed known in the full wavefront sensing FOV defined by a diameter of a circle. The PAOLA s GLAO modes, combined with the freedom to place the DM and SHWFS at a specific conjugated height, are used for our S-GLAO performance simulations Performances with DGS Wavefront Sensing In this section, we will calculate the DGS-based S-GLAO performances at average and good seeing conditions at the NIR J and H bands. We first calculate the S-GLAO performance at average seeing condition with Fried parameter r 0 ¼ 0:07 m and an isoplanatic angle of 4 with four layers of turbulence. These two parameters are evaluated at 0:55 μm only, if without any special statement. During our simulations at different seeing conditions and wavelengths, the DM pitch and the SHWFS subaperture diameter are fixed as 0.16 m, which matches the Fried parameter at the 1:25-μm J band. We assume that our S-GLAO works at the NIR J and H bands with a 1.6-m aperture telescope, which is the largest solar telescope currently available. Table 2 shows the simulation results of the S-GLAO Strehl ratio at the NIR J band as a function of off-axis field angle (i.e., as a function of imaging FOV), corresponding to different optimized FOV (OFOV) sizes that are used for wavefront sensing. For comparison, we also list the performance of the conventional adaptive optics (CAO), which uses only one on-axis discrete guide star for wave-front sensing. At average seeing condition, CAO delivers Strehl ratios of 0.78 and 0.12 at on-axis (0 imaging FOV) and 60 imaging FOV, respectively. For the S-GLAO, at 40 OFOV, it delivers a Strehl ratio of 0.19 at 60 imaging FOV, which is better than the 0.12 of the CAO, at the cost of reducing the on-axis Strehl ratio, which is Again, at 60 OFOV, the S-GLAO delivers a Strehl ratio of 0.22 at the 60 imaging FOV. In general, the S-GLAO with a large OFOV should deliver a smooth Strehl ratio in the imaging FOV. The S-GLAO performance can be further improved by slightly shifting the DM above the telescope aperture, while the SHWFS is also conjugated to the same height. For the four-layer turbulence profile, the best conjugate height is 400 m, which is less than the conjugate height h with 50% vignetting for the marginal ray of the 60 OFOV. According to the criterion discussed in 2.1, the vignetting introduced on the SHWFS is negligible. Table 3 shows the Strehl ratios with the DM and SHWFS conjugated to 400 m height, in which all specifications are identical with that of Table 2, except for the conjugated height. The plots of Strehl ratio as a function TABLE 6 STREHL RATIOS AT DIFFERENT OFOVS AT1:65-μmWAVELENGTH AND GOOD SEEING CONDITIONS (r 0 ¼ 91 mm), WITH A 400-M CONJUGATE HEIGHT SR: CAO SR: 40 OFOV SR: 60 OFOV

7 SOLAR GROUND-LAYER ADAPTIVE OPTICS 475 TABLE 7 STREHL RATIOS WITH DIFFERENT OFOVS AT1:25-μm WAVELENGTH AND AVERAGE SEEING CONDITION (r 0 ¼ 70 mm), WITH 400-M CONJUGATION HEIGHT SR: CAO SR: 40 OFOV SR: 60 OFOV of the off-axis FOV position are shown in Figure 4. Now, for the 40 OFOV, the S-GLAO delivers a Strehl ratio of 0.58 and 0.25, on the on-axis and 60 imaging FOV, respectively, which is a significant improvement for the performances on the off-axis imaging FOV. For the 60 OFOV, the S-GLAO delivers a Strehl ratio of 0.48 and 0.28 on the on-axis and 60 imaging FOV, respectively. Since the best conjugate height differences are very small at different seeing conditions, in our further discussions, we will assume that both the DM and WFS are always conjugated at the altitude of 400 m. At good seeing conditions with Fried parameter r 0 ¼ 91 mm and 6 isoplanatic angle, both the CAO and S-GLAO performances improve significantly, as shown in Table 4. Now, for the 40 OFOV, the S-GLAO delivers a Strehl ratio of 0.74 and 0.42, on on-axis and 60 imaging FOV, respectively, and the performances in view of Strehl ratio are excellent for both the on-axis and the off-axis positions on the imaging FOV. For the 60 OFOV, the S-GLAO also delivers excellent performances. This suggests that both 40 and 60 OFOVs are optimal choices for the S-GLAO wave-front sensing. If we increase the wavelength, the performance of the CAO and the S-GLAO will improve accordingly. Table 5 shows the Strehl ratios at the 1:65-μm wavelength and average seeing conditions, while the corresponding plots of Strehl ratio as a function of the off-axis FOV position are shown in Figure 5. The 40 OFOV S-GLAO delivers a Strehl ratio of 0.73 and 0.43, respectively, on on-axis and 60 imaging FOV, while at the 60 OFOV, the S-GLAO now delivers a Strehl ratio of 0.65 and 0.46 on onaxis and 60 imaging FOV, respectively. As estimated, the S-GLAO performances significantly improve at good seeing condition at the 1:65-μm wavelength. AsshowninTable6,theS-GLAOwitha40 /60 OFOV can deliver excellent performances over the entire 120 imaging FOV, with a Strehl ratio better than 0.34/ Performances with LGS Wavefront Sensing It is interesting to know the LGS-based S-GLAO performance, in comparison to that of the DGS system. In this section, we will calculate the LGS-based S-GLAO performances at average and good seeing conditions at the NIR J and H bands; this is done by using PAOLA s GLAO full modes, in which the large wavefront sensing FOV (i.e., OFOV) is defined by a FOV diameter of 40 and 60, respectively. Table 7 shows the LGS-based S-GLAO Strehl ratios, at 1:25-μm NIR J-band average seeing conditions. The associate plots of Strehl ratios are shown in Figure 6. For the 40 /60 OFOV, the S-GLAO delivers good performance in the imaging FOV between 0 60, with a Strehl ratio between 0:24 0:68=0:27 0:58. Comparing Table 7 with Table 3, it is clear that the LGS wavefront sensing system delivers a better Strehl ratio at a small imaging FOV close to the on-axis position, while both LGS and DGS systems deliver almost the same performances at an off-axis position with a large imaging FOV. The extra gain on the on-axis Strehl ratio for the LGS system results from the large continuous FOV for the wavefront sensing. At good seeing, the S-GLAO system delivers better performances, as estimated. Table 8 shows the LGS-based S-GLAO Strehl ratios, at 1:25-μm NIR J-band good seeing conditions. Now, the S-GLAO delivers excellent Strehl ratios in the 0 60 imaging FOV, with a Strehl ratio between 0:41 0:81=0:44 0:74, for the 40 /60 OFOV. Even at 120 imaging FOV, the S-GLAO will still be able to deliver a good Strehl ratio in the order of 0:18 0:20. Again, comparing Table 8 and Table 4, Strehl ratio FOV (arc second) FIG.6. Strehl ratios of LGS-based system with different OFOVs at 1:25 μm NIR at average seeing conditions. Solid line represents the CAO; dotted line, optimized for 40 OFOV; dash-dotted line, optimized for 60 OFOV. See the electronic edition of the PASP for a color version of this figure.

8 476 REN ET AL. TABLE 8 STREHL RATIOS AT DIFFERENT OFOVS AT1:25-μmWAVELENGTH AND GOOD SEEING CONDITIONS (r 0 ¼ 91 mm), WITH A 400-M CONJUGATE HEIGHT SR: CAO SR: 40 OFOV SR: 60 OFOV it still holds true that the LGS wavefront sensing system delivers a better Strehl ratio at a small imaging FOV close to the on-axis position, while both LGS and DGS systems deliver almost the same performances at a large imaging FOV. At the 1:65-μm H band, the S-GLAO system delivers excellent performances. Table 9 shows the LGS-based S- GLAO Strehl ratios, at 1:65-μm H-band average seeing conditions. The corresponding plots of Strehl ratios are shown in Figure 7. Now, the S-GLAO delivers a Strehl ratio between 0:42 0:80=0:45 0:73 in the 0 60 imaging FOV, for the 40 /60 OFOV. At the large image FOV of 120, the S-GLAO delivers a Strehl ratio in the order of 0:20 0:22. Again, compared to the corresponding DGS system, the LGS system provides an extra gain at small imaging FOV. Table 10 shows the LGS-based S-GLAO Strehl ratios, at the 1:65-μm H-band good seeing conditions. With the improvement of the seeing conditions, the S-GLAO deliver excellent performances with a Strehl ratio between 0:34 0:89=0:37 0:84 in the imaging FOV, for the 40 /60 OFOV. One should note that while the S-GLAO can still provide better performance at a imaging FOV equal to or larger than 24, the CAO also be able to provide a good performance in the image FOV, which is consistent with the past AO experiments, in which solar high-resolution images are occasionally available at excellent seeing condition over a large FOV, even without AO correction at some good sites. 3. DISCUSSION A challenge for the discussed S-GLAO system is the highspeed wavefront sensing and computation power requirements over a large field of view. Solar AO uses a SHWFS, in which the local vector of each subaperture is solved by the calculation of cross-correlation over a two-dimensional field of view. Depending on the seeing conditions, current AO systems use FOV for wave-front sensing, which is sampled by pixels in each SHWFS subaperture. Since no subpixel accuracy is applied (limited by AO running speed), this yields a sampling accuracy of 0:5 =pixel. This sampling approach provides enough precision for a small wavefront sensing FOV such as However, for a large wavefront sensing FOV such as FOV, the pixel-samplingaccuracy will provide poor sampling performance, which is not enough to lock on a solar structure for wavefront sensing, and in this case an AO system will not be able to work properly (i.e., the AO closed loop may crash). For example, if a sunspot of is sampled by pixels, it results in a sample accuracy of 2 =pixel, and an AO system cannot work properly because of the poor precision of wavefront sensing. This is the case that has been confirmed by the past GLAO experience with the SNO DST (Rimmel et al. 2010). We found that in this case, the AO system can only work properly when subpixel accuracy is available, and this can explain why the solar GLAO experience with the NSO GLAO was not successful. While maintaining a fast open-loop running speed over 1000 Hz, our software, developed with LabVIEW, is optimized for subpixel sampling accuracy with a precision better than 1/ 10 pixels, which allows us to sample a large wavefront sensing FOV with enough accuracy. For example, with FOV, sampling by pixels in each WFS subaperture, our SWFAO can deliver a sampling accuracy of 0:2 =pixel, which is good enough for our S-GLAO system, and this is confirmed by the good performance delivered by our current portable AO that uses a wavefront sensing FOV of (Ren & Dong 2012; Ren & Zhu 2013; Ren et al. 2014b). Camera-link format CMOS cameras are used for solar wavefront sensors by current AO systems, which can deliver TABLE 9 STREHL RATIOS AT DIFFERENT OFOVS AT1:65-μm WAVELENGTH AND AVERAGE SEEING CONDITIONS (r 0 ¼ 70 mm), WITH A 400-M CONJUGATE HEIGHT SR: CAO SR: 40 OFOV SR: 60 OFOV

9 SOLAR GROUND-LAYER ADAPTIVE OPTICS Strehl ratio FOV (arc second) FIG. 7. The same as Fig. 6, but at the 1.65 NIR wavelength. See the electronic edition of the PASP for a color version of this figure. image data at a rate up to 850 MB=s. Recently, commercial CoaXPress interface CMOS cameras are available, which enable the transfer of an ever growing data stream with 25 GBit=s. For example, the EoSens 3CXP CoaXPress interface camera manufactured by Mikrotron can grab image at a speed of up to 4500 fame per second (fps) at a region of pixels. These latest CMOS camera technologies, combined with our dedicated subpixel wavefront sensing software, make our S-GLAO system immediately applicable with current largest solar telescopes. Figure 8 shows our recent S-GLAO result of the sunspot 2139 H band images with the AO-off (left) and AO-on (right) on the 2014 observation run with the 0.7-m NSO DST. In this test, the SHWFS has 9 9 subapertures. The entire sunspot is used as a guide star for wave-front sensing, and a wavefront sensing FOV up to is used, which is sampled by pixels in each subaperture. Combined with the latest commercial computer with 2 Intel Xeon CPUs (in which each CPU has 12 cores), our current S-GLAO achieves an open-loop speed over 2000 Hz, which is sufficient for the NIR imaging with the largest solar telescopes. FIG. 8. NIR H band images of sunspot 2139 with AO off (left) and AO on (right). 4. CONCLUSIONS In this publication, we present two different wavefront sensing approaches for S-GLAO high-resolution imaging over a large FOV; the LGS uses a large-continuous region as a guide star, while the DGS uses several discrete guide stars, each defined by a small region. The simulations indicated that LGS-based S-GLAO has a better performance in the imaging FOV around the on-axis position, while at a large imaging FOV, the performance for both wavefront approaches is comparable. The simulations also indicated that both the 40 and 60 OFOV can deliver good performances; however, the 60 OFOV has the advantage of accommodating a large as well as a small sunspot for wavefront sensing. An important feature of the S-GLAO is that it does not need to reconstruct the wavefronts tomographically, and is much simpler to implement. The S-GLAO, with the DM and WFS pair conjugated slightly above the telescope aperture, can effectively correct the dominant ground-layer turbulence. Our simulations demonstrated that the S-GLAO can deliver good performances for high-resolution imaging over a large imaging FOV up to 60 in the NIR J and H bands, at average and good seeing conditions. Our wavefront sensing techniques can be applied to any one-dm AO systems and should be considered for the future upgrades of these existing conventional AO systems. TABLE 10 STREHL RATIOS AT DIFFERENT OFOVS AT1:65-μmWAVELENGTH AND GOOD SEEING CONDITIONS (r 0 ¼ 91 mm), WITH A 400-M CONJUGATE HEIGHT SR: CAO SR: 40 OFOV SR: 60 OFOV

10 478 REN ET AL. The LGS-based S-GLAO can use an entire sunspot for wavefront sensing, which avoids the problem of isoplanatic angle for sunspot imaging, and is currently the only adaptive optics that provides such a unique solution to this problem. Other AO systems, including future solar MCAO, cannot effectively address this issue, since they need to use nearby granulation patterns for wavefront sensing, which will result in a performance degradation for sunspot imaging. This work is supported by the National Science Foundation (NSF) under the grant ATM , and partially supported by the National Natural Science Foundation of China (NSFC) under the grants , , , , and , as well as CAS special funding KT We thank Dr. Aglaé Kellerer helping to correct the writing and providing valuable comments. We acknowledge the anonymous referee s comments and work which significantly improved the quality of our manuscript for this publication. Our S-GLAO systems were tested with the National Solar Observation telescopes at both Kitt Peak and Sacramento Peak. The National Solar Observatory is operated by the Association of Universities for Research in Astronomy under a cooperative agreement with the National Science Foundation, for the benefit of the astronomical community. REFERENCES Andersen, D. R., Stoesz, J., Morris, S., Lloyd-Hart, M., Crampton, D., Butterley, T., Ellerbroek, B., Jolissaint, L., et al. 2006, PASP, 118, 1574 Berkefeld, T., & Soltau, D. 2010, Astron. Nachr., 331, 640 Ellerbroek, B. L., Gilles, L., & Vogel, C. R. 2003, Appl. Opt., 24, 4811 Grossmann-Doerth, U., Keller, C. U., & Schüssler, M. 1996, A&A, 315, 610 Johnston, D. C., & Welsh, B. M. 1994, J. Opt. Soc. Am. A, 11, 394 Jolissaint, L. 2010, J. European Opt. Soc., 5, Keller, C. U. 2005, in ESO Symposia: Science with Adaptive Optics, Adaptive Optics Observations of the Sun (New York: Springer), Kellerer, A. 2012, Appl. Opt., 51, , Appl. Opt., 53, 7643 Kellerer, A., Gorceix, N., Marino, J., Cao, W., & Goode, P. R. 2012, A&A, 542, A 2 Khomenko, W. V., Collados, M., Solanki, S. K., Lagg, A., & Trujillo- Bueno, J. 2003, A&A, 408, Lin, H., & Rimmele, T. 1999, Astrophys. J., 514, Marino, J., & Wöger, F. 2014, Appl. Opt., 53, Ragazzoni, R., Marchetti, E., & Rigaut, F. 1999, A&A, 342, L 53 Ragazzoni, R., Farinato, J., & Marchetti, E. 2000, Proc. SPIE, 4007, 1076 Ramsauer, J., Solanki, S. K., & Biémont, E. 1995, A&AS, 113, Ren, D., & Dong, B. 2012, Opt. Eng., 51, Ren, D., & Zhu, Y. 2013, in Adaptive Optics Progress, A Solar Adaptive Optics System, ed. R. K. Tyson, Ren, D., Zhu, Y., Zhang, X., Dou, J., & Zhao, G. 2014a, Appl. Opt., 53, 1683 Ren, D., Li, R., Zhang, X., Dou, J., Zhu, Y., & Zhao, G. 2014b, Proc. SPIE, 9148, W Rimmele, T. R., Woeger, F., Marino, J., Richards, K., Hegwer, S., Berkefeld, T., Soltau, D., Schmidt, D., et al. 2010, Proc. SPIE, 7736, Rimmele, T. R., & Marino, J. 2011, Living Rev. Solar Phys., 8, 1 Schmidt, D., Gorceix, N., Zhang, X., Marino, J., Coulter, R., Shumko, S., Goode, P., Rimmele, T. R., et al. 2014, Proc. SPIE, 9148, U Socas-Navarro, H., & Sánchez Almeida, J. 2002, ApJ, 565, 1323 Socas-Navarro, H., Beckers, J., Brandt, P., Briggs, J., Brown, T., Brown, W., Collados, M., Denkere, C., et al. 2005, PASP, 117, Tokovinin, A., & Viard, E. J. 2001, Opt. Soc. Am. A, A18, 873