HI surveys:

In short, the points to clarify are: 

- the A/Tsys values of the European design: the 10500 is below the
  strawman design to which all original plans and calculations were
  based. To get the same sensitivity a four- fold integration time is
  needed. This is not a problem if the instrument has multiple FOV
  capabilities, so dedicating a beam for a long time to HI surveys 
  will not impair other observing programs.

- The FOV of the Karst design is too small, but here a focal plane
  array could help. We should say this so the designers know. This
  may then be a level 2-3 compliance.

- HI surveys: the 50% baseline area is too large in the European design.
  This can change in the design. Moreover it is important how the 50%
  collecting area is distributed within the 100 km! So this requirement
  in my opinion is level 3.

HI Surveys:

   China: FOV far too small. Lowest frequency prohibits observing HI 
          beyond z = 3.7 which is too low a redshift.
  Europe: A/Tsys is too low but multiple beams compensate that. Number
          of short spacings may not be ideal, but further study needed to
          set final requirement.
   India: baseline distribution needs check 
  Ozlens: baseline distribution needs check 
   Ozcyl: baseline distribution needs check 
     USA: baseline distribution needs check 
  Canada: baseline distribution needs check 


Source: Thijs van der Hulst, Chair of WG4



     China:  system FOV far too small, and the required 200-element FPA
             elements may be tough to fit on the focal plane (at 150
             MHz especially). The frequency range goes down only to
             300 MHz, so only z = 0 - 3.7 can be studied. This is too
             low: since we know that there are objects at z > 6, we
             must be able to look for them in HI
     Europe: combination of A and number of beams OK, though the
             baseline range is non-ideal (needs shorter baselines,
             which should be manageable).
     India:  looks pretty good, except that the lower frequency limit
             needs extending. The baselines should be available (I
             have no detailed indormation on layouts.)
     Ozlens: should be fine, though the shortest baseline population
             may be non-optimal
     Ozcyl:  again fine except for baseline population.
     USA:    fine except for lower frequency bound, which may be too
             high for some of the science (going only to z = 8.5 may
             be too much to give up, although it covers the
             presently-known population of galaxies)
     Canada: should do the job, and the baseline range could be
             available

Source: Mark Birkinshaw from Working Group 4


SR amended China to a 2 for consistency with other working groups.