First Sagittarius A* Event Horizon Telescope Results. I. The Shadow of the Supermassive Black Hole in the Center of the Milky Way

In this section we summarize the EHT observations and data processing for Sgr A∗. We refer the reader to for a comprehensive discussion of the EHT instrument, for details on the 2017 observing campaign and data processing, and for additional details specific to the Sgr A∗ data processing.

In 2017, the EHT observed Sgr A∗ on five nights between April 5 and 11 using an array with 8 participating observatories (see Figure 1). Weather conditions were good or excellent at all sites on all five observing nights. The most sensitive element in the array, the Atacama Large Millimeter/submillimeter Array (ALMA), only observed Sgr A∗ on April 6, 7, and 11; the pv (pv) only observed Sgr A∗ on April 7. In this initial series of papers, we focus on the observation with the best baseline coverage: April 7. In addition, we utilize the April 6 observations for testing, validation, and selected multi-day analyses. Our multi-wavelength coverage indicated that there was an X-ray flare on April 11, accompanied by increased 1.3 mm variability; we will consider that more complex data set in future work.

Each site, except the JCMT and ALMA, received data in two circular polarizations simultaneously. The JCMT received a single circular polarization, and ALMA received orthogonal linear polarizations that are converted to a circular basis in post processing. All EHT sites recorded data in two frequency bands at 227.1 and 229.1 GHz, referred to as the “low” and “high” bands, respectively. The total recording rate at each fully-outfitted station is 32gigabits/second. The data were written to arrays of hard drives at each site, that were then brought from all sites to common locations where we computed the complex cross-correlation in the electric fields measured for each pair of stations.

Following the initial computation of these correlations, residual phase and bandpass errors were corrected with two independent processing pipelines, EHT-HOPS and rPICARD. The data were then a priori calibrated using the system equivalent flux densities (SEFDs) for each telescope. Multiplication by the geometric mean of the SEFDs of the two stations on a given baseline converts dimensionless correlation coefficients to flux densities. SEFDs ranged from 60 Jy at ALMA to 5×10⁴ Jy at low elevation at SMT. Further corrections are still required, as some sites do not measure the SEFD continuously, and in any case the SEFD does not capture many amplitude corrupting effects such as pointing and focus errors. A “network calibration” process, which uses redundancies in the array (e.g., co-located telescopes) to provide more accurate time-variable gain normalization for sites with co-located partners, was performed. Sgr A∗ presents a special case for calibration, as it varies significantly in time, and is surrounded by extended emission that corrupts the visibility amplitude for baselines within local arrays like ALMA and SMA. Wielgus et al. [120] discusses the techniques used to estimate the time-resolved flux density of Sgr A∗ for this calibration. For the remaining stations, gain corrections were computed using observations of the calibrator targets J1924-2914 and NRAO 530.

Figure 2shows the EHT baseline coverage and interferometric visibility amplitudes (“visibilities”) for Sgr A∗. The longest baselines have an interferometric fringe spacing of 1/|u→|≈24μas, which defines the diffraction-limited angular resolution of the EHT. The visibility amplitudes have two deep minima, the first at |u→|≈3.0Gλ and the second at |u→|≈6.5Gλ. The amplitudes have a baseline dependence that is similar to that of an infinitesimally thin ring with a 54μas diameter that has been blurred with a circular Gaussian kernel with 23μas full width at half maximum (FWHM). The ring diameter is primarily constrained by the minima locations, while the width is determined by the amplitude of the secondary visibility peak between the minima. For instance, to be consistent with both a first minimum falling between 2.5−3.5Gλ and a second minimum falling between 6−7Gλ, the ring must have a diameter of ∼50−60μas.

However, the EHT data show evidence for complex and asymmetric source structure beyond a simple ring model, including a strong dependence on visibility amplitudes with the baseline position angle (especially near the first visibility minimum) and in the closure phases measured on triangles of EHT baselines, which differ significantly from 0° and 180°. The EHT visibility amplitudes also show indications of residual calibration errors. To assess and quantify the source morphology of Sgr A∗ more generally, we applied a variety of imaging and modeling methods using both the interferometric visibility amplitude and phase information (section V). These methods included a variety of approaches to account for residual calibration errors, including iterative self-calibration, simultaneous fitting of a model and the residual station gains, and analyses that used only closure quantities.