This transformation modelled some, but not all, of the distortions in NZGD1949. Therefore D coordinates may be out of terms with surrounding NZGD1949 marks of other orders.
The phenomenon of D coordinates in the geodetic database being "out of terms" with the non-D coordinates is not a new problem, it is just easier to detect now than in the past. The reason for this issue being raised now is due to the re-survey of a significant part of the existing NZGD1949 network as part of the development of NZGD2000.
Historically, when a conventional third order survey was undertaken all third order stations in the region were re-observed and subsequently re-adjusted. Inconsistencies between coordinates would only be evident when observing between third order marks that were part of different regional networks.
If the third order regional network covered a standard traverse network, then it would not be expected that the measured bearing and distance between an urban mark and the third order mark to agree with the computed bearing and distance until the standard traverse network had been adjusted into the third order regional network.
The problem with the D coordinates is highlighted with fourth order coordinates because the distance between stations with new and historical coordinates is relatively small and they are interspersed (See Figure 1).
In Figure 1 all of the stations are existing marks with NZGD1949 coordinates. The new survey only occupied the solid circle marks and therefore no new coordinates were generated for the marks with open squares. The effect of the new survey adjustment is that all the 5 solid circle marks have two sets of coordinates.
The problem of a third survey tying into a mixture of old and new coordinates is shown in Figure 2.
In Figure 2 positions 1 and 3 are for the same physical ground mark (likewise positions 2 and 5) but they have historical and current coordinates which differ. The bearing and distance between positions 1 and 2 will not be the same as between positions 3 and 5. Both sets of coordinates are correct.
The relative coordinate accuracy between the new 'd' type coordinates is high and the same is true between the historical coordinates. The problem of coordinate inconsistency is only revealed when the two types are mixed. For a cadastral survey the origin of bearings needs to be computed from station coordinates which were determined via the same adjustment network.