212man - I have assumed accurate positioning at the DR/MAPt in order to provide the diagram; this always produces less manoeuvring for the divergent procedure (if attempting to regain the into-wind track) for obvious reasons.
The work on the revised ARA obviously took account of the FAA system which provided fixed tracks - this was also the basis of Norwegian modification (provide to eliminate the turning missed approach in a tight obstacle environment). Such procedures are probably too restrictive as they rarely provide more than a small number of fixed paths for any location. On the plus side, the offset distance can be calculated with some accuracy and meet international standards (DOC 8168).
I must say I was surprised when discovering that the revised procedure would be omni-directional; however this does introduce issues as, unless there is accurate measurement and representation of each platform's environment in the data-base, it will have to rely upon a single distance based upon the worst case (for each location). This is one effect of an offset procedure with a standard (straight ahead) missed approach (but deterministic and safer than a turning missed approach).
HC - I arrived at the same conclusion yesterday when drawing the diagram.
The main hazards of any non-precision procedure are not in the instrument segments but during the level segment and at the DR/MAPt (that is why there has been a move towards CDFA - see my earlier post on this page - to the point where non-CDFA procedures are penalised in the approach minina under the latest revisions of EU-OPS). It has been established for a number of years that the most critical point of any procedure occurs at the MAPt and the studies (CAA and FSF ALAR) provide substantial evidence of of this.
It is noteworthy that the studies only considered straight-in approaches; the ARA has the added difficulty that, in addition to acquiring the target, the pilot has to further manoeuvre the aircraft to the landing phase. Whatever profile is produced, it will not alter the basis of the procedure that there has to be an offset because the obstacle to be avoided in the missed approach is also the landing site.
As has been recognised for some time, once visual reference (in accordance with the requirements) has been established, landing into wind will require an S-turn to regain the into-wind track followed by a deceleration and let down. If the DR/MAPt is at 0.75nm, this will not result in a tight manoeuvre (less for the 'divergent' than the 'offset'); if at 0.5nm the whole procedure becomes squeezed because of the more acute angle and the reduced space in which the manoeuvre is performed (and this on top of the additional complexity of acquisition of a target 30 degrees to one side).
The revised ARA procedure appears to take an 'input' of the proposed track (we assume that this will be the into-wind track) and then constructs the complete profile (vertical and horizontal). There is no reason then why the profile could not be constructed such that the DR/MAPt is downwind of the landing site. This would also assist in the acquisition of the target as the drift angle would put the target closer to the centre of the windscreen. Such a procedure would then only require a single and maximum turn of 30 degrees (less if the DR/MAPt is at 0.75nm) to put the aircraft on the final track.
Regardless of any of this, the radar will still be necessary to ensure that there are no obstacles in the approach path - this might be complicated with any complex approach path computations. I believe that there are assumptions that the radar will not be necessary and other methods can be used; I do not subscribe to that contention for the same reason that I believe that use of the 'E' function of the EGPWS is flawed offshore.
Jim