Speckle interferometry, as a technique for high-resolution optical imaging, began in 1970 with the work of Labeyrie (1970), using ideas for fast imaging to remove atmospheric effects. Labeyrie (1970) showed that taking short exposures removed the effects of seeing-induced fluctuations that were causing distortions in the wavefront from a distant star. Removing these distortions allowed the diffraction limit of the telescope to be reached.

Speckle cameras initially used photographic plates, but continued with the various detectors of the day, such as photomultiplier tubes, video tubes, and reticons (e.g., Bonneau and Foy, 1980; McAlister et al., 1987; Weigelt and Baier, 1985; Horch et al., 1992; Balega et al., 1993). Initial studies mainly focused on bright binary stars using 1- to 4-m telescopes and repeated imaging to produce precise stellar orbits (e.g., McAlister et al., 1989). Significant advances in astronomical detectors, in particular the more quantum efficient charge-coupled devices (CCDs), provided the next leap forward in this field, allowing for photon intensification, digital outputs, and higher signal-to-noise (S/N) observations to be obtained (e.g., McAlister et al., 1989; Mason et al., 1997; Hartkopf et al., 2000; Horch et al., 2000). In recent years, the introduction of the electron-multiplying CCD (EMCCD) as a detector (Tokovinin and Cantarutti, 2008; Horch et al., 2011) has been a game-changer. With ultra-fast readout, essentially zero read noise, near-perfect Quantum efficiency (QE), optical flatness, and ease of use, EMCCDs have revolutionized the field of speckle imaging. Fourier-based data reduction and image reconstructions using autocorrelation and power spectra techniques (Horch et al., 2012; Horch et al., 2015) are enhanced using the bispectrum technique developed by Weigelt (1977) and Lohmann et al. (1983) allowing phase information to be determined that resolves the ambiguity present in autocorrelation analysis (e.g., Howell et al., 2011; Hope et al., 2022).

This renaissance in speckle imaging has led to the development of new dedicated speckle instruments (e.g., Maksimov et al., 2009; Tokovinin et al., 2010; Howell et al., 2021b; Clark et al., 2020; Pedichini et al., 2016) placed on some of the largest telescopes in the world. A summary of former and current speckle imagers in astronomy is presented in Scott et al. (2021). Speckle imaging is no longer limited to bright star astrometry; it has expanded into many areas of point source and non-point source imagery (e.g., Salinas et al., 2020; Scott et al., 2021; Shara et al., 2022). Fainter astronomical targets can now be observed (Howell et al., 2021a), the overall data quality and S/N ratio of the observations are greater, and the final reconstructed images have more fidelity (e.g., Howell and Furlan, 2022).

In contrast to infrared (IR) adaptive optics (AO) systems, optical speckle imaging on 8-m-class telescopes routinely achieves an inner working angle (IWA) at the diffraction limit of the telescope (20–30 mas across the optical Lester et al., 2021), uses far less expensive instrumentation, and does not require a (laser) guide star, enabling higher observational efficiency. In speckle interferometry, the detection contrast is proportional to 1/ ( s e e i n g ) 2 but the angular resolution is unaffected ( λ /D). This is in contrast to normal imaging and IR/AO in which the final image resolution is affected by the native seeing, being proportional to λ / r 0 r 0 ∼ native seeing, Fried, 1966). Speckle imaging performed in the optical bandpass (350–1,000 nm) provides the highest angular resolution available today on any single telescope, delivering ∼ 4 times better angular resolution than IR AO observations in the K -band.

This paper presents a summary of the first 8 years of astronomical imaging observations using the highest resolution, deepest contrast speckle instruments available, ‘Alopeke and Zorro, which are mounted on the twin 8.1-m Gemini North and South telescopes in Hawai’i and Chile (Scott et al., 2021; Howell and Furlan, 2022). We review the major advances and scientific areas covered with these instruments since their introduction in 2018.

‘Alopeke and Zorro (the ‘ o ̄ lelo Hawai’i and Spanish words for “fox”) are identical instruments that use iXon Ultra 888 EMCCD cameras to provide simultaneous speckle imaging in two optical band passes, yielding high-resolution reconstructed images of the observed source. SDSS u , g , r , i , z , H α , and 4 narrow-band filters are available in two filter wheels passing light dichorically split at 700 nm. These imagers are used with one of two circular fields-of-view: speckle mode (diameter = 6.7 arcseconds) or wide-field mode (diameter = 60 arcseconds). The left panel of Figure 1 shows ‘Alopeke with the covers removed, revealing the tightly packed innards that contain two filter wheels, the optical elements, and the two ANDOR EMCCD cameras extending from the box. The right panel of Figure 1 shows Zorro in its permanent mount location at the GCAL port, attached underneath the Gemini South 8.1-m primary mirror. The small space available for the speckle instruments did not allow for an atmospheric dispersion corrector (ADC) to be included. However, this exclusion is offset by the large advantage that our visitor instruments have in that they are permanently mounted on Gemini and therefore always available to use. We discuss our “no ADC” mitigation strategy below. A complete description of these two visiting Gemini instruments, including relative transmission as a function of wavelength in the 350–1,000 nm range, can be found in Scott et al. (2021) and at the Gemini Observatory instrument web pages ¹ .

‘Alopeke and Zorro proposal demand at Gemini varies semester by semester, but on average it is 5%–10%, exceeding the Gemini South Adaptive Optics Imager (GSAOI) demand and nearing the demand for other visiting instruments (e.g., IGRINS), or even facility instruments (e.g., GNIRS). Any user can request time on these instruments using the regular Call for Proposals, or Director’s Discretionary Time (DDT) when needed. Principal Investigators (PIs) from Gemini partner countries can also propose for ‘Alopeke and Zorro time using Fast Turnaround (FT) proposals ² .

Speckle instrumentation is fairly simple, inexpensive, and small compared to other instruments mounted on large-aperture telescopes. ‘Alopeke and Zorro are roughly the size of a carry-on suitcase. The most difficult aspect of speckle observing is the quantity and short duration of the exposures, which is easily managed by the fast read-out capabilities of the EMCCDs. Using the Gemini mirror coating reflectivity, the optics and filter transmissions, and the QE of the EMCCDs, optimal speckle observations can be carried out between 400 and 940 nm where the total throughput is 40% or higher (Scott et al., 2021). As a mitigation for the lack of an ADC, typical speckle observations are performed at an airmass of 1.4–1.5 or lower (elevations above 45°) and narrow band filters are used. Speckle imaging at Gemini is often done under (bright) moonlit sky conditions, for which the narrow band filters are preferred as well. The EMCCDs can also function as normal CCDs, allowing traditional, long exposure digital images to be obtained.

Speckle imaging of a target requires many thousands of short exposures (10–60 ms in length) to be obtained and processed. This large number of images is required to build up sufficient S/N, especially at contrasts greater than four or five magnitudes ( ∼ 1 0 − 2 ) , at very close angular separations, and/or for fainter targets. Although this large number of exposures may seem daunting, at tens of milliseconds per image, typical speckle observations last only a few minutes per target (Hope et al., 2019; Howell and Furlan, 2022). Stars with V magnitudes from 1 to ∼ 12 require only ∼ 5 min of observation time, during which 3,000 to 5,000 thousand exposures are collected. However, targets as faint as R = 19 can be observed by using ∼ 50 min of on-source time. In a typical night, 40 to 50 sources are observed. Scott et al. (2021) and Howell and Furlan (2022) provide full details on the relation between the magnitude of the source, the filter, the brightness of the target and the sky conditions to the total integration time at the source and the resulting S/N. To provide the reader with a gauge for the time required for a variety of observation types, we list in Section 5, the total on-source times and filters used for each observation.

Almost all currently used speckle image reconstruction software packages are based on Fourier speckle interferometric methods (e.g., Labeyrie, 1970; Lohmann et al., 1983; Horch et al., 2001; Horch et al., 2015). Speckle imaging collects a highly magnified image of a source in a very short time period. Each collected image consists of numerous “speckles” that cover the area of the point spread function. A single speckle exposure therefore has little total flux, thus the need to obtain many individual short exposures, each sampling the atmospheric turbulence. Simply averaging the many exposures would yield just a native seeing image. Thus, the time averaged power spectrum is produced consisting of individual power spectra plus a transfer function of the combined telescope-atmosphere system. The transfer function can be evaluated by a separate measurement of a single point source (i.e., a PSF standard) allowing a solution for the target’s power spectrum. Although this formulation can yield amplitude information on multiple sources or extended flux structures such as angular separations or magnitude differences, reconstructed images cannot be produced. To do so, the triple correlation or bi-spectrum technique is employed (Lohmann et al., 1983). A complete description of the speckle imaging process from observation through image reconstruction is beyond the scope of this short review paper, but is provided in detail in Beavers et al. (1989). The results presented in this paper, and all fully reduced data in the archives, are based on our implementation of the data reduction methods as described in Howell et al. (2011) and Horch et al. (2012). Our standard data reduction pipeline (Horch et al., 2012; Howell et al., 2011) provides robust 5 σ magnitude contrast limits on stellar companion or circumstellar material detections (e.g., Howell et al., 2016). Speckle reconstructions scale the output image maximum value to 1.0 at the center of the image. All of the images presented in this paper (except Nova V906 Car and Eros which are shown using just a red intensity color map) have their brightest pixel scaled to 1.0 and are presented on a min-max log scale using the relative intensity colorbar shown in Figure 3.

Fourier speckle deconvolution techniques are computationally efficient but, as described above, require a PSF model power spectrum usually obtained via routine observation of stars taken from the HR or HD catalogs that are known or assumed to be single based on past spectroscopic or imaging observations. These PSF standards are assigned to each target by the observing team prior to the scheduled observations. They are observed near in time to the targets and with similar sky locations. Each PSF standard observation requires an additional ∼ 3 minutes of observing time. Modern image reconstruction methods that are based on blind deconvolution techniques and can reach deeper contrasts plus provide more accurate astrophysical results are beginning to be implemented for speckle image reconstructions (e.g., Howell et al., 2024). Figure 2 presents a typical Gemini speckle imaging result for a point source (the star TOI-5873), and the discovery of a very close stellar companion that is 1.6 magnitudes fainter than the primary. Note the 180-degree autocorrelation quadrant ambiguity that occurs when stars have comparable brightness (Horch et al., 2012). During data reduction, a “best determination”, using all sets of images taken of the target plus bispectrum analysis, is made to assign the correct location. The astrometric precision for binary stars observed at Gemini yields stellar separations to ± 1 mas and ± 1° in position angle (Lester et al., 2021). Photometric magnitude differences for stars in multiple systems typically have uncertainties of ± 0.25 magnitudes, increasing by ∼ 2 × for very close or wide ( > 0.9) pairs (e.g., Howell et al., 2011; Horch et al., 2012). Raw and fully reduced archival data from Zorro and ‘Alopeke can be found in the Gemini Observatory (Hirst and Cardenes, 2017) ³ or Exoplanet Follow-up Observing Program ⁴ archives.

In this section, we discuss various scientific targets that have been observed using ‘Alopeke and/or Zorro. Most of these observations were obtained as test cases or engineering studies (unless otherwise noted) and as such, many have not been published previously. These studies allowed researchers to assess the potential for using speckle imaging to accomplish their scientific goals, and have formed the basis for a number of stand-alone proposals targeting the detailed study of single objects, or larger sample collection to enable more global results.

The ‘Alopeke and Zorro speckle cameras are permanent visitor instruments at the Gemini North and South telescopes in Hawai’i and Chile. One can request to use these instruments for general queue proposals, DDT and FT proposals, Long and Large Proposals, and Target of Opportunity proposals as well. Note that the oversubscription rate of the Gemini telescopes hovers near 3, as compared to 6–7 for HST, Chandra, and JWST.

Optical speckle imaging has the advantage of darker skies (than the IR), no need for AO natural or laser guide stars, inexpensive and simple instrumentation, and easy setup and use. Gemini speckle observations can be made from 350 nm to 1,000 nm, providing the ability to gather a spectral energy distribution across the optical bandpass for any detected companion. Additionally, speckle imaging reaches the diffraction limit and is indifferent to whether the target star is single or multiple; the observations and data reduction processes are uniformly applied regardless of the target’s multiplicity. This is not an ability shared by coronagraphic instruments.

The applications of speckle imaging have broadened greatly since ‘Alopeke and Zorro were first installed at Gemini Observatory. While first used to observe stars at high angular resolution to search for stellar companions, speckle imaging is now employed to decipher the shape of Solar System bodies, study the outflows and ejected material of supernova, observe extragalactic eruptive events, and investigate other exciting astrophysical phenomena.

As the world of ground-based observational astronomy moves into the era of 30–40 m-class telescopes, speckle imagers should be one of the first facility instruments available. Speckle imagers are inexpensive, simple to construct, small in format, light in weight, and easy to operate. The use of EMCCDs and their high-speed imaging ability not only allows speckle imaging to be accomplished but fast time-series photometric observations as well. Such time-series observations of bright stars can also find telescope engineering applications in the analysis of x,y centroids over time to examine guider and tracking errors, and unwanted telescope vibration modes. This type of engineering study has already been successfully implimented at Gemini using ’Alopeke and Zorro. The high-speed imaging cameras can also be used as telescope guiders or wavefront sensors, such as has been done at the WIYN telescope for use with the NEID spectrograph (see, e.g., Gupta et al., 2021).

Speckle instrument imagers can also function as standard CCD imagers. Providing, as we have in ’Alopeke and Zorro, an optional wider field of view, the EMCCD cameras can be used as conventional CCD imagers, providing long-exposure photometric images. As a “first-light” instrument, obtaining images throughout the larger final field of view would yield optical quality checks of the delivered PSF across the focal plane.

As speckle interferometry moves to these larger telescopes, reaching the optical diffraction limit on a 30-m-class telescope would provide, at 400 nm, an spectacular ground-based angular resolution of 4 mas. Such imaging resolution would be fantastic, allowing a multitude of new astrophysical science cases to be explored.