Comments on 'Plan for an Enhanced ALMA' presented 12 June 2000 by Japanese members of the ASAC. I have incorporated comments from presentations by S. Guilloteau and Koh-Ichiro Morita at the ALMA Technical Workshop, 17 Feb 2000 in Tokyo, the ASAC meeting in Leiden, and other venues. I. Antennas: A. Additional 12m antennas As Stephane noted, increasing the number of 12m antennas from 64 to 78 results in an ALMA in which sensitivity per partner is conserved, and overall sensitivity is increased. This seems like a reasonable goal. B. The Terahertz Array A compact array of seven 6-8m antennas is added, with each antenna bringing an equal value to that of one 12m antenna. The exact number and size are poorly determined. An alternative which poses sizable problems for pointing and surface accuracy might be to build one submillimeter telescope of 24m diameter. Scientific Merits of a Compact Array: I believe that the compact array brings a very good enhancement to ALMA as it promises production of science otherwise not addressed by ALMA, at the highest frequency windows. The compact array could operate at supraTerahertz frequencies for the 15% of the time at which transparencies in the supraTerahertz windows exceeds 30%, for instance. As Guilloteau noted, this Terahertz Array could operate independently of, or in addition to the main array, with the antennas on the latter underilluminated to improve efficiency, field of view, and effective pointing performance. At 1 THz, the resolution could be 3-4" over a 10" field of view (FOV) for an 8m telescope, somewhat larger for 6m telescopes (but see below), operating as a stand-alone array. Issues affecting number and diameter of antennas includes a. As antennas become smaller, i.e. 6m, it may be difficult to accommodate a standard suite of receivers, and standard cabin, compromising the economy of scale. As the array becomes more specialized, it becomes more difficult to operate and maintain. b. As number of antennas increases, it becomes more difficult to pack them together to obtain the smallest baselines. The array need not be concentric with the large array, but should perhaps be located nearby for operation on medium baselines with underilluminated 12m antennas. The number of antennas is limited to perhaps 16. 16 x 6m antennas has the same collecting area as one 24m antenna. c. As antennas become larger, it is easier to accommodate the standard suite of receivers. As the compact array may be operated as a standalone Terahertz Array, and may require additional receivers to operate at supraTHz frequencies, it seems unwise to restrict the receiver cabin. Some space may be recovered if the THz Array need not operate at the longest wavelengths. This suggests perhaps 8m antennas. However, existing implementations of inhomogeneous arrays have employed antennas with diameters varying by a factor of two. Nine 8m telescopes have the same collecting area as one 24m antenna. Seven 8m antennas provides adequate sensitivity for matching with an ALMA mosaicking 25% of its observing time. A. Imaging quality should be very much improved, through provision of a. short baseline information (6-10m) for imaging with the larger array. Although simulations are just beginning, ALMA is expected to proved excellent imaging at millimeter wavelengths, with some degree of quality loss at submillimeter frequencies. Especially at high frequencies, direct measurement of 8m spacings provides much better accuracy than recovery of this information through combination of interferometric plus on-the-fly mapping data. b. improved cross calibration between interferometer and single antenna performance c. improved performance for the combined array, particularly in bands 9 and 10. d. For a main array performing wide field imaging 25% of available time, the Terahertz Array spends essentially all its time collecting short spacing information for the wide field imaging experiments. This will pose some potential logistical problems--different atmospheric conditions, calibrator fluxes, scheduling conflicts but this difficulty is probably only at the annoyance level. e. The compact array may be constructed in a fixed array, with long baselines provided through baselines with underilluminated 12m antennas. Comments on 'Plan for an Enhanced ALMA' presented 12 June 2000 by Japanese members of the ASAC. I have incorporated comments from presentations by S. Guilloteau and Koh-Ichiro Morita at the ALMA Technical Workshop, 17 Feb 2000 in Tokyo, the ASAC meeting in Leiden, and other venues. I. Antennas: A. Additional 12m antennas As Stephane noted, increasing the number of 12m antennas from 64 to 78 results in an ALMA in which sensitivity per partner is conserved, and overall sensitivity is increased. This seems like a reasonable goal. B. The Terahertz Array A compact array of seven 6-8m antennas is added, with each antenna bringing an equal value to that of one 12m antenna. The exact number and size are poorly determined. An alternative which poses sizable problems for pointing and surface accuracy might be to build one submillimeter telescope of 24m diameter. Scientific Merits of a Compact Array: I believe that the compact array brings a very good enhancement to ALMA as it promises production of science otherwise not addressed by ALMA, at the highest frequency windows. The compact array could operate at supraTerahertz frequencies for the 15% of the time at which transparencies in the supraTerahertz windows exceeds 30%, for instance. As Guilloteau noted, this Terahertz Array could operate independently of, or in addition to the main array, with the antennas on the latter underilluminated to improve efficiency, field of view, and effective pointing performance. At 1 THz, the resolution could be 3-4" over a 10" field of view (FOV) for an 8m telescope, somewhat larger for 6m telescopes (but see below), operating as a stand-alone array. Issues affecting number and diameter of antennas includes a. As antennas become smaller, i.e. 6m, it may be difficult to accommodate a standard suite of receivers, and standard cabin, compromising the economy of scale. As the array becomes more specialized, it becomes more difficult to operate and maintain. b. As number of antennas increases, it becomes more difficult to pack them together to obtain the smallest baselines. The array need not be concentric with the large array, but should perhaps be located nearby for operation on medium baselines with underilluminated 12m antennas. The number of antennas is limited to perhaps 16. 16 x 6m antennas has the same collecting area as one 24m antenna. c. As antennas become larger, it is easier to accommodate the standard suite of receivers. As the compact array may be operated as a standalone Terahertz Array, and may require additional receivers to operate at supraTHz frequencies, it seems unwise to restrict the receiver cabin. Some space may be recovered if the THz Array need not operate at the longest wavelengths. This suggests perhaps 8m antennas. However, existing implementations of inhomogeneous arrays have employed antennas with diameters varying by a factor of two. Nine 8m telescopes have the same collecting area as one 24m antenna. Seven 8m antennas provides adequate sensitivity for matching with an ALMA mosaicking 25% of its observing time. A. Imaging quality should be very much improved, through provision of a. short baseline information (6-10m) for imaging with the larger array. Although simulations are just beginning, ALMA is expected to proved excellent imaging at millimeter wavelengths, with some degree of quality loss at submillimeter frequencies. Especially at high frequencies, direct measurement of 8m spacings provides much better accuracy than recovery of this information through combination of interferometric plus on-the-fly mapping data. b. improved cross calibration between interferometer and single antenna performance c. improved performance for the combined array, particularly in bands 9 and 10. d. For a main array performing wide field imaging 25% of available time, the Terahertz Array spends essentially all its time collecting short spacing information for the wide field imaging experiments. This will pose some potential logistical problems--different atmospheric conditions, calibrator fluxes, scheduling conflicts but this difficulty is probably only at the annoyance level. e. The compact array may be constructed in a fixed array, with long baselines provided through baselines with underilluminated 12m antennas. f. Radford notes: Should nearby mountain tops be condsidered for the location of a teraherz/super-terahertz array? Radiosonde profiles show the bulk of the water vapor is trapped below an inversion layer fairly close to the ground at least some of the time (e. g., flight 85, 1999 November 7 UT 4). Putting a high frequency array 400-700 m higher, i. e., on Cordon Honar (5400 m), Cerro Chascon (5650 m), or Cerro Sairecabur (5750 m), would lift it above the inversion layer and dramatically improve the observing conditions. This is not a novel idea -- it was suggested at the Cornell workshop last week, among other venues. At the very least, we should determine from the existing radiosonde data the expected improvement in observing time. And more radiosonde launches certainly would help. I note that, on the other hand, baselines to the ALMA array would not be particularly useful in most cases if the THz Array were sited far from it; cross calibration might suffer if the atmosphere at the THz Array site were very different from that at the main ALMA site. B. Science with the Terahertz Array To some extent, the compact array would bring the most identifiable new capability to ALMA. However, it is not clear that the technology is yet ready for construction of a Terahertz Array. However, that technology is being developed for the FIRST mission (2007), and may be available during ALMA construction. a. Frequency. The frequency extent addressed by ALMA would be extended to include the three supraTHz windows accessible from the Earth's surface with reasonable transparency. 1. 1-1.06 THz window. Recently, the HHT detected CO 9-8 in this window, which also contains the CS 21-20 line. Transmission is about 20% when transmission in the 450 micron window is about 70%. This will be referred to as the 1.03 THz window. 2. 1.25 - 1.37 THz window. This window is somewhat diced by atmospheric lines, rather like the Band 8 window, but contains the CO 11-10 line and reaches a transparency nearly that of the 1.03 THz window mentioned above. Of the three supraTHz windows, this is probably of least interest. 3. 1.5 THz window. This window contains the lower frequency of the two [N II] lines, the CO 13-12 transition, and the HCN 17-16 line. Owing to the uniqueness of the [N II] line as a probe of the ISM, and its strength in the Milky Way (measured by COBE) and other galaxies (estimated from ISO measurements of the higher frequency line in other galaxies using simple CLOUDY models), this is the primary interest of the three supraTHz windows. Transmission reaches 20% but the window is somewhat broader than the 1.03 THz window. 4. Dust continuum emission increases from most objects through these windows, and would provide an interesting target. Spectral baseline, for determination of spectral energy distributions, for example, suggests that the 1.5 THz window is of most interest. 5. [C II] emission at 2.2 THz enters the 1.5 THz window at z=0.47, providing a window on this line in galaxies at a time between the present epoch and that of peak star formation. 6. However, receivers to cover these bands would come at some future time when technology improves. In the interim, receivers using this future technology might be tested on the THz array antennas. Enhanced frequency coverage would not be part of ALMA construction. b. Sensitivity - 1. Sensitivity can be calculated for the proposed 7 x 8m array. These antennas should achieve better surface accuracy than the 12m antennas; an assumption of 10 microns might be reasonable, better might be achievable. This accuracy should provide an efficiency near 60%. With more baselines, and a diameter better targeted to filling the gap in spacings available to ALMA, a 16 x 6m array sensitivity is also calculated. 2. Goals of a THz Array, and a compact array operating to provide short spacings for ALMA at high frequencies, conflict somewhat in that both goals can only be achieved in the best weather. 3. Estimates - Proposed 7 x 8m array. For operation in the 1.5THz band, a receiver achieving 25 h nu/k is assumed (SSB). 20% zenith transmission through the atmosphere received by an 8m antenna with 55% efficiency and a main beam size of 7" are also assumed for a source at 1.3 AM. In one minute, the array could achieve 0.3 Jy sensitivity, of .01K in brightness temperature. In a single channel 1 km/s wide, this is 12 Jy km/s, or 0.5K with shortest baseline 1.5D=12m. 4. Estimates - Proposed 16 x 6m array. For operation in the 1.5THz band, a receiver achieving 25 h nu/k is assumed (SSB). 20% zenith transmission through the atmosphere received by an 6m antenna with 55% efficiency and a main beam size of 9" are also assumed for a source at 1.3 AM. In one minute, the array could achieve 0.23 Jy sensitivity, of .006K in brightness temperature. In a single channel 1 km/s wide, this is 9 Jy km/s, or 0.2K on baselines of 1.5D=9m. c. Implementation 1. Japan has proposed providing these antennas over the period FY2002-2008. See Nakai's comments; my statement is incorrect. 2. Some correlator redesign would be needed for an ALMA with an increased number of antennas. Escoffier has offered the opinion that the basic design could accommodate 88 antennas, sufficient for 78 12m antennas plus 10 smaller ones. However, since the THz Array antennas may not often be operated with the 12m antennas, some additional capacity may be available. II. Receivers and LO A. Receiver Bands The ASAC voiced interest in Band 4 (2mm) since CO at redshifts of 0.45 to 0.8 for the 2-1 line will fall in this range. Band 3 covers redshifts of 0. to 0.35 or so, depending upon its lower frequency, for the 1-0 line. Thus these two lines give good sensitivity to redshifted CO for the lowest excitation lines over the redshift range in which most evolution in the star formation rate has occurred, if one is to believe published estimates. For higher redshifts, the compaction of the spectrum provides readier access to a range of redshifts. I believe that this should be a priority also. It would be useful to lower the lower edge of band 3, in my opinion, from 86 GHz to 84 GHz (which allows complete overlap with the band covered by the 3mm VLBA receivers). This would have to reach 80 GHz for complete coverage of the two lowest CO lines for redshifts between 0.35 and 0.45. The weather at Chajnantor is not always submillimeter; I believe that Band 4 should be a priority, in place of Band 10, which I believe would benefit by waiting for the technology to mature. B. LO This was not really developed in the memo of 12 June. Does this include expansion of the photonic LO system to frequencies below 300 GHz? III. Correlator (from 12 June note plus Okamura presentation 17 Feb 2000) A. The proposal is that NRAO provide the first quarter of the correlator now planned, to be replaced by a second generation correlator producing 125 kch/IF from 85 antennas. Architecture to be determined by EU/Japan. Advantages cited, especially by Okamura in February for FX design, include: a. Increased number of baselines addressed b. Increased sensitivity by increasing the number of bits (not specified); easier with FX c. Increased bandwidths: advantages 1. Simultaneous wide bandwidths and high spectral resolution Advantages: multi line imaging not restricted by having only n IFs 2. Imaging line surveys possible with resolution matched to source 3. Wide velocity coverage with high resolution may be important where a range of conditions are encountered, such as protostellar disks with warm kinematically active centers and cold exteriors with, e.g. slow infall. 4. At 1.5 THz, 2 GHz corresponds to 400 km/s, suggesting the existing design might place some restrictions on the THz Array. 5. Wideband high resolution imaging is useful for studies of radio loud quasars such as 3C84. High accuracy continuum subtraction needed 6. Recombination line studies suggest broad coverage at good resolution 7. Absorption line studies of distant galaxies 8. Serendipity, such as water masers in NGC4258. d. Disadvantages of increased bandwidths 1. Huge throughput may mean additional computing costs. These should be assessed. 2. FX design can result in higher development costs 3. More cabling needed for FX design. See Nakai's emails on further ideas on this design.