Tuesday, 11 July 2017

ProjectDev: Ensuring correct distances on maps

When it comes to rail maps, we have things like stations that have distances associated with them, and marking the correct distances is a pretty important facet of designing maps.

When the Quail Atlas came out in the third and fourth editions, it was metricated from the first two editions which were produced in the pre-metric era and had imperial distances shown. I would guess most of the imperial data came from earlier working timetables (two in particular were re-released by NZRLS for the North and South Island dating from the early 1950s and are important reference works that I may yet want to access for the maps other than Otago Central). 

The metric Quails apparently chose to convert all the imperial distances to kilometres which is something I disagree with most strongly. I don't believe they were obliged to convert all maps to the metric measurements and this brought with it other issues. Basically the imperial distances are a kind of "namespace" if you like. That particular namespace was defined by the process by which these measurements were obtained and the era in which that occurred. It is extremely possible that every railway line has its own distance measurement namespace and that there may in fact be several different distance measurement namespaces for any particular line.

What is important to recognise is that the measurement namespaces for a line like the Otago Central were obtained by separate measurement processes and the problem with simply converting the imperial distances from another namespace to the metric distances risks confusing these distances with the metric measurement namespace that NZR obtained for all of their lines that were open at the time of metrication in 1974. The metric measurements were probably obtained by track evaluation car whereas some of the imperial measurements came from hand measurement of the lines as they were built (chain pegs and mileposts were installed as each section of line was constructed). The reason that the metric measurements often differ from metricated imperial measurements is that inaccuracies in the earlier measurements could occur and that lines were also improved, deviated and curves straightened or eased changing the actual length of the line.

It's important to recognise that the measurements from one namespace therefore will often differ from the measurements in another namespace that have been metricated (as is the case in the Quail Atlas for all lines and stations that closed before metrication). The metrication of NZR also resulted in all existing bridges and tunnels being renumbered into new namespaces which has created similar problems with the potential for confusion and duplication of numbers across the different (imperial and metric) namespaces.

Essentially it is my desire to have the correct distances marked and to ensure there is a clear understanding and avoid confusion the following policy has been adopted:
  • All metric distances are those which can be obtained from an official NZR / KRL working timetable. 
  • Where a metric distance cannot be obtained from the above sources, the distance is displayed in imperial measurements (decimal miles).
Where I am having to use the Quail Atlas as a reference source I am converting metric measurements back to imperial, for the present. But I hope to get copies of the working timetables that NZRLS produced in order to have the correct distances shown in imperial units and also some that the Quail atlas is missing (possibly stations that closed before the NZRLS's working timetable reprints were released).

The Otago Central maps have mileposts marked from the chainage charts. There has never been any intention to metricate the milepost markings as the actual km posts on the ground are unlikely to align with positions obtained from converting a position on a chainage chart (for example, 161 km = 100 miles 5 chains).

Now a footnote to the last post: Unfortunate issues with the latest master of Qgis 2.99 means I can't use it for map editing and I have had to revert to an older version running on my Windows computer. Just to make things even harder there are networking issues with my mainpc running Xubuntu 17.10 that means none of the virtual machines that I could run on it can connect to the shared volume on it. But the Windows computer can. So at the moment I am trying to determine with the Xubuntu developers why this networking issue exists. It's taken a couple of days I didn't really have to get an older build of 2.99 running on the Windows computer to ensure I have something that actually works for all the rest of the map work I am going to do until the project finishes at the end of the year.

Well I have now sorted the networking issue but still have to use the Windows computer along with an older development master running on a Linux computer because of various issues. But we will get there.