The scientific capabilities required in the U. S. MMA were refined in community science workshops sponsored by the NRAO throughout the decade of the 1980s and confirmed by the September 1995 MMA Science Workshop held in Tucson, AZ. Five reports were written following the Tucson Workshop that summarize the science goals in the following categories:

1. Cosmology and Extragalactic
2. Star Formation and Stellar Evolution
3. Galactic Molecular Clouds and Astrochemistry
4. Solar System
5. Sun and Stellar

The scientific requirements for the LMSA were similarly discussed and summarized in a document NAME.

In November 1998 the MAC decided to increase the diameter of the primary to 10m, as negotiations continued to merge the MMA and LMSA, which had 15m primary diameter.  In December 1999, the MMA Oversight Committee recommended a further increase of the MMA diameter to 12m.  Other characteristics of the antenna specifications, notably its pointing specifications, were tightened by the ratio of the increase of diameter.

The two projects merged into the ALMA project in 1999 June, resulting in a merger of the similar science needs. Bids for construction of the prototype antennas were received and contracted at the specifications developed for the MMA, scaled to the new diameter.  Contractors believe that they can meet these specifications.

Early discussions in the U. S. had contemplated complementing the MMA observations with an array of smaller antennas, but for 8m primary diameter it was shown that this was not necessary. In November 1998 the MAC discussed an inhomogeneous array but the primary diameter was increased mainly because of simplicity of an array with a uniform design. There was general realization that wide-field imaging at submillimeter wavelengths would not be optimal but detailed studies were not done to quantify this owing to time pressure.
 

II. Science Drivers for an Atacama Compact Array

1. Submillimeter imaging

For the 12m antenna the field of view in the submillimeter becomes small, on the order of 6" at 900 GHz. The pointing specification is 0".6, or 0.1 beamwidth at this frequency. The efficiency is expected to drop to 30% or so. These parameters are reminiscent of the operation of the VLA at 7mm--imaging can be done but it will not be precision imaging and images will in general only cover the central region of the primary beam.

For a 7m antenna, such as might populate the ACA, the primary beam would be 10" with 1/16 beamwidth pointing. The smaller antenna might be built to better accuracy, perhaps with a ten micron surface (Baars, ALMA Memo No. 339 ), resulting in 70% aperture efficiency. Although detailed simulations remain to be made, it is clear that the ACA alone would produce more precise images. If one can match sensitivity appropriately with ALMA's main array the ACA offers some hope of producing precision submillimeter images also. Furthermore the major science goal of ALMA, to achieve wide field imaging, might be achievable in the submillimeter, where IRAS and SCUBA images have revealed that the sky is festooned with structure.

2. Recovery of short spacings

In ALMA imaging, spatial frequencies between 6 and 10m are not well recovered during standard mosaicking operations. At all frequencies, the ACA would contribute spacings within this range if the primary diameter lies within this range (see Guilloteau, 'Single-Dish and Short Baselines' 11 July 2000).

3. SupraTHz operations The ACA might also be used as a dedicated array for observations in the supraTHz windows (1.1, 1.3 and 1.5 THz) where the site provides acceptable atmospheric transparency. Tippers in these windows (details?) are now operational at Chajnantor and estimates of the amount of usable time in these windows should be available soon. In a fixed configuration, this array might provide resolutions of 2-4" in the higher frequency bands. However,
 

III. Character of the Atacama Compact Array

If the primary goal is to improve submillimeter performance of ALMA, then the ACA may be operated for longer periods for a given imaging experiment than might ALMA alone. The question of how many antennas of which size has received attention in a number of memos. Guilloteau (2000) compared expected sensitivities in flux and brightness of the compactly configured ALMA to that of a single antenna and of an ACA. He noted that the compact ALMA can only have a certain number of short baselines from geometric considerations. Matching ACA performance over the same range of short spacings suggests that the ACA can have a modest size. He showed that considering brightness sensitivity only an ACA of 7-8m antennas operating with 4 dedicated ALMA antennas equipped with nutators could provide sensitivity to appropriately fill the gap in spatial frequencies. Considering calibration sensitivity considerations, he recommended an ACA of 12 8m antennas. Welch (2001, ALMA Memo No. 354), emphasizing the deleterious effects of pointing errors upon attempts to estimate intermediate spacing information from on-the-fly sampling, suggested the ACA could improve estimation of this missing information. He compared point source sensitivity of an ACA matching that of the shortest spacings on the main ALMA array to recommend that the ACA consist of about a dozen antennas of about 6m size.
 

IV. Simulations Performed to Date

Yun (2001 ALMA Memo No. 368) prepared simulations for the ACA based on arrays of 19 x 6m antennas and 13 x 8m antennas. He simulated a mosaic of an H$\alpha$ image of M51, modified slightly to have a size of 2' square at 345 GHz, using AIPS routines VTESS (employing a maximum entropy algorithm) and UTESS (employing a maximum emptiness algorithm) for the ACA alone and the ACA with ALMA. Yun also optimized the ACA configuration using te Kogan sidelobe minimization routines. He found little difference between arrays employing 6m antennas to provide missing short spacing information and those employing 8m antennas for this task. The importance of total power data is confirmed, as is the merit of the homogeneous array concept. However, Yun notes that the mosaic pointings did not go far enough off target to define the zero level, i.e. establish a 'guard band' to constrain the algorithms.

Morita (2001 ALMA Memo No. 374) also studied wide field imaging with an ALMA which included an ACA. A maximum entropy routine within the SDE software package was employed along with a four component pointing error model. His results showed that ALMA-alone antenna pointing errors pose a serious problem for creating high quality images owing to errors in the visibilities in the central poorly sampled (u,v) region. Dynamic range and fidelity index for the deconvolved image decline from (>1000, >50) at 3mm to (300~400, <10) at 630 GHz in these 12m-only simulations. Seven to eighteen smaller (6 and 8m) antennas, circularly arrayed, are shown to compensate for this, giving slightly better results for 6m antennas among the eight array designs used. ALMA-ACA cross correlations and ACA total power data was not included; the ACA was assumed to follow the same pointing model and errors as ALMA. A 'guard zone' of blank sky data was included to guide the data through deconvolution as this was found to be important. Total power data (12m antennas) was convolved with a beam equivalent to that of the 6m antenna to suppress phase errors originating in pointing of the 12m antennas. With pointing errors and thermal noise, 230 GHz dynamic range was improved from 250 to ~320 to ~330 for ALMA plus arrays of 6m antennas. Fidelity index was improved from ~10 to ~17-~23. Similar improvements could be obtained by improving pointing performance of the 12m antennas by a factor of 2.

Gueth, Guilloteau and Pety, in a draft memo on ALMA+ACA Simulation (http://iraux2.iram.fr/~alma/pointing-only.pdf), considered simulations including noise and pointing errors using a CLEAN-based method. The input was the M51 HST image used for configuration studies, as a snapshot of a 7-field hexagonal mosaic without adjacent blank sky. A 12 antenna ACA of 7m antennas with a north-south stretched random configuration was used in addition to ALMA in its compact configuration. Additionally, four 12m antennas provided total power data. Without noise and without pointing errors, these simulations show marked improvement when ACA observations are added.

A number of 230 GHz simulations were performed with various input parameters. Pointing errors are comprised of wind-induced errors, themally induced errors and a combination of the two. For the ACA, the rms pointing error was taken to be the same as the 12m but with different distribution: measurement errors dominate the ACA while wind and thermal errors are smaller. The results show that the median fidelities are improved by 50-100% by adding the ACA. The improvement increases when the median fidelity is computed on the pixels with highest intensities. A possible operations scheme for ALMA with the ACA was suggested, for a model in which ALMA requires the short spacing data 25% of the time. During all its time, the ACA and four 12m antennas measure short spacings for that 25% of ALMA projects by time. Observations are synchronized such that four 12m antennas may be used with the ACA for calibration.
 

V. Recommendations

The report of the Configuration PDR suggested: "10) The ACA could consist of about 12 to 16 7-m antennas, with a close packing ratio of 1.25, plus 4 12-m antennas used for calibration and single dish work. ALMA configuration design should take into account an area of about 55-m in each dimension, where this ACA will be located. This area should be near the compact configuration of ALMA, and as such may interfere with the most compact "intermediate" configurations. The ACA configuration should be designed for best brightness sensitivity. Since shadowing constraints will play a significant role, it is allowed to have at most 50 % extra pads to provide good UV coverage even at the extreme declinations. However, since simulations have shown the ACA configuration plays a role in the final image quality, proposed designs should be compared using available simulation tools.