ALMA-US Strawman Configuration Designs

(Toronto URSI Workshop, Sept. 1999)

What Drives the Array Design?

Science:

• Resolution: One of the prime scientific goals is the detection and resolution of extra-solar planets and protostellar systems, either by direct imaging or by accurate astrometry. Subarcsecond imaging capability is a key scientific driver for ALMA, and a configuration with baselines of 10 km or longer is essential.
• Sensitivity: ALMA should deliver an order of magnitude improvement in sensitivity over all existing millimeter arrays. Excellent brightness sensitivity is highly desirable capability, and a close packed configuration is also required. A robust response of the configurations to tapering is another desirable quality since tapering of data may be unavoidable for multi-transition or multi-frequency analysis.
• Mosaic and Short Spacing: The homogeneous array concept requires that the antennas be fairly closely packed (1.3D) in order to be able to measure spatial frequencies in the range of the antenna diameter (see Cornwell 1988, A&A, 202, 316). Since mosaic imaging is expected to be used a common and a popular observing mode for ALMA, the configuration design should include a provision for excellent short spacing data. An added concern over shadowing needs to be addressed, either incorporating an elevation restriction or by an N-S elongation.
• Image Fidelity & Dynamic Range: High image fidelity is essential for multi-transition studies and continuum spectral index studies. High dynamic range is needed for most imaging studies, particularly for imaging bright objects such as planets. An array design that produces a well behaved synthesized beam and thus allows successful deconvolution is highly desirable for achieving this goal. The synethesized beam need not necessarily be a Gaussian, however.
• Fourier Plane Coverage: The large number of array elements for ALMA (e.g. 64) ensure an excellent instantaneous Fourier Plane (FP) coverage in general. Unlike the existing arrays, the image quality is then limited by the extrapolation in the gaps of the FP coverage rather than by the crude sampling of the visibilities. A ring-like configuration yields a nearly uniform Fourier plane distribution and thus offers a highly desirable characteristic (see Woody 1999, MMA Memo 270). It also offers the best resolution of sources near the resolution limit. The synthesized beam has large close-in sidelobes, but the sidelobe suppression algorithm developed by L. Kogan is shown to be highly effective (see below).

Operation:

• Minimizing Pads, Cabling, & Road: Requring maximum re-use of pads and cabling for the purpose of minimizing the construction cost can have a strong influence in the array design. Should the imaging capability and sensitivity of the array be compromised to reduce the cost of construction and maintenance? If so, what is the acceptable level? For the purpose of this strawman array design exercise, some efforts are made to minimize these costs but only after determining which types of array designs offer the desired imaging and sensitivity characteristics. This can become a potentially important issue and may need to be reconsidered if the cost of each pad runs as high as $100K.
• Ease and Speed of Reconfiguration: Should the imaging capability and sensitivity of the array be compromised in favor of the ease and speed of reconfiguration? If so, what is the acceptable level? Related issues are: (1) can we afford to station a full time, permanent reconfiguration crew at 16000 ft; (2) how can we minimize the number of hours lost during reconfiguration and subsequent calibration; and (3) how can we maximize the utility of the limited manpower?
• Dynamic Scheduling: ALMA be able to reconfigure quickly to take advantage of climatic changes. The VLA-like set configurations do not allow such flexibility. A telescoping array also requires N(antenna)/N(mover per day) days to reconfigure -- over two months if there are 64 antennas and only one antenna can move each day! Further, a telescoping array introduces an additional layer of complication to the scheduling of observations as well as reconfiguration.
• Array Efficiency: Can we afford to trade the scientific output of the array with the ease of the array operation? Set configurations are easier to plan and execute, but a greater loss of overall efficiency may occur by requiring more data to be tapered (see below).
• Redundancy: Some redundancy can be useful for cross-calibration. On the other hand, this causes sidelobes to pile up at particular locations. It might be better to distribute the error instead. Are the trade-offs sufficient to warrant building in some redundancy in the design?

Cost-Benefit Analysis (MMA Memo 199 & 265)

Kogan Optimization (MMA Memo 171)

Keto (1997, ApJ, 475, 843) and others have previously argued for array designs that optimizes the FP coverage. Such a design makes a good sense for the arrays with small numbers of elements (e.g. <10) since the image quality is limited by how well the visibility can be measured with the limited sampling of the Fourier plane. When the numbers of the array elements are as large as 64, then even the instantaneous FP coverage should be very good, and a different optimization critiria may be more important.

Kogan (1997, MMA Memo 171) suggests an alternative array optimization scheme which minimize the sidelobe response within the certain numbers of synthesized beam radii. By minimizing the sidelobe responses, this optimization aims to allow the best possible performance for any deconvolution methods, such as CLEAN or MEM. See the memo for the full mathematical desciption and how this algorithm is implemented in practice.

Using a mathematical proof, Kogan (1999, PASP, 111, 510) also demonstrates that the negative sidelobes of an array's synthesized beam in a snapshot observation depends only on the number of the elements. For a 64 element array, the maximum negative sidelobe level should be about 1.6% for a snapshot.

Competing Reconfiguration Concepts

Set Configurations:

Continuous Reconfiguration:

==> consider a hybrid scheme?

Practical Concerns

Site:

• Topography: Cerro Chajnantor, Cerro Chascon, as well as the quebradas and local gradient limit where the pads can be placed. In general, all concentric configuration with larger than 3 km diameter will have difficulty getting around these topographical obstacles (see below).
• Gas Pipeline: Similar to above but man-made. The three crossing points are already negotiated.
• Science Reserve Boundary: Not a hard limit as it could potentially be changed with some effort.

Logistics:

• Post-move Calibration: measuring delays, pointing, baselines, etc. This could have an important impact on overall array efficiency as some of these measurements require 3-6 antennas at once and take more than 2-3 hrs (see Yun & Kogan 1999, MMA Memo 265). While a post-correction is possible for any baseline errors, an antenna is useless until its delays and pointing model are determined. This also has a serious impact on the real time data pipeline processing.
• Cost of Additional Configuration: The cost of one additional configuration is approximately $5-10M, comparable to the cost of 1 or 2 antennas. Trading the cost of one or two antennas for an additional configuration may be easily justifiable if the array capability can be significantly improved.
• Cost of a Transporter: Estimated to be about $1M each. Should this impact the array design? Regardless the type of intermediate configuration design we adopt, the reconfiguration to and back from the "10 km" array will require enough transporters to carry this out in a reasonably short time. Reconfiguring from the 3 km configuration to the 10 km configuration will take approximately 3.5 hr per antenna assuming an average one way travel distance of 15 km. To complete this reconfiguration in 6 days, we will need at least 5 transporters if each transporter can move 2 antennas per day (8 days with 4 transporters). If we had only 2 transporters, this reconfiguration will take 16 days!
• Weather: Wind (and snow/rain) may offer a major challenge to any reconfiguration plan (see Holdaway & Owen 1995, MMA Memo 147). The site characterization data suggests a strong diurnal variation in wind speed (see Holdaway et al. 1995, MMA Memo 159), and a windy condition is common in the afternoons.
• Cycling Time: How often should we return to a particular configuration? At least once around the full configuration cycle each year, but a phase shift needs to be incorporated to allow for the celestial seasons. The largest configuration may only need to be visited every other time.

Strawman Arrays

Compact Array:

• Surface Brightness Sensitivity
• Mosaic Imaging and Short Spacing Information

Intermediate Arrays:

• Double Rings with a scale factor between 2 and 4
• Short spacing data in each configuration
• Less concentrated for good FP coverage and angular resolution

"10 km" Array:

• Highest angular resolution
• Uniform FP coverage

---

A Strawman Layout at the Site

A layout for a possble set of ALMA configurations are shown below. This drawing is the output of an exercise only meant to demonstrate how a potential ALMA configurations may be matched to the site topography and other restrictions. At the moment, only limited topographical information sufficient for 3 km array is available, but additional information should become available shortly. An important outcome of this exercise is that there are limited options for a 500 meter diameter flat area at the site, which makes fitting any concentric or spiral type configuration difficult. The 10 km configuration will most certainly have to be independent of the more compact configurations.

---

Remaining Tasks

Imaging Study (Simulations & Evaluations)

• Mosaic imaging
• FP coverage: sidelobes and deconvolution
• Snapshot versus earth rotation synthesis

Site Study

• Location of the compact array (which determines how the remaining arrays may or may not fit into the site constraints)
• Largest configuration allowable and its geometry with respect to the topography
• Soil study (e.g. much cheaper to put pads on bedrocks)

Detailed Optimizations

• N-S extension & Hybrid Arrays
• Operational details. For example, some 3-6 antennas (5-10%) may be unavailable at all times either due to failures or for reconfiguration.

 

---

Last modified September 15, 1999

myun@nrao.edu