Friday, 27 January 2017

Otago Central Railway [6A]: Relocated station buiildings

Oops, I had better put in a couple of pictures of the railway stations in their new locations.

The second photo was found here:
The title refers to the means used to get this building up a hill, which you can read about in the web site, as well as the general area.

Otago Central Railway [6]: Location maps for relocated Ida Valley and Kokonga stations

Following on from my previous post it has been interesting and useful to locate the Kokonga and Ida Valley station buildings which were taken intact from the railway after they closed. I am not sure of the dates in which this occurred but the buildings are publicly accessible today and are used as musterer's huts and in one case can also be booked by the public as a tramping hut.

The main benefit I get from locating them is to be able to have the right size building shown on the maps. The secondary interest will be for other people who want to go and take pictures of them or whatever.

Here is the overview map of where the two buildings are located relative to the railway line. The border between Canterbury and Otago is also shown, so you can see that the Ida Valley building is actually in Canterbury.

If we now add some roads to the picture, we can start to narrow in on the locations of the two buildings.
In this image we see lower centre the township of Naseby and we can see that the Kokonga Station building is situated on Mt Buster Road. It is at the end of this road, where tramping tracks begin. It is 5 km south of the regional boundary, so only just in Otago.
The Ida Valley building doesn't have road access; instead it is on a tramping track. There are various ways to access the track system. It is about 15 km north of the Canterbury-Otago boundary, so inside Canterbury.

Thursday, 26 January 2017

Otago Central Railway [5]: Hyde-Kokonga

So in the last week I have been pushing ahead with the maps and have now reached milepeg 65, Kokonga station. Right now I am drawing a station layout for that yard which had a few sidings including a ballast pit and a coal loading stage. Part of the interesting aspect of this is that I found photos of the old railway station building, which is offsite. It turns out the Kokonga and Ida Valley station buildings were taken off the site and shifted into the Ida range in South Canterbury (the border between Canterbury and Otago is actually quite close to Wedderburn) and I am adding these locations to the map.

Hyde station.

Hyde Township.

Prices Creek.


Part of Kokonga (still being drawn).

Friday, 20 January 2017

Otago Central Railway [4]: Mileposts

Today I decided I was going to put mileposts in. This is possible because of the detail on the chainage charts. First issue to resolve is how to display them. In the OSStations table I have currently set up mileposts with a choice of either km or as units. This means miles can be decimalised as miles and chains, which is like displaying hours and minutes as a decimal, because the base system for a chain is not the decimal system, just like minutes is not decimal. The number of chains in a mile is 80 chains, not 100. So I have decided to create a third possible unit, M, which is going to be decimalised miles. In other words, if chains were the basis of this, I would have to convert them to a decimal mile before entering their value. Whilst we may recognise it is confusing for non technical people to understand that a decimal point doesn't mean what it should in this context.

So in starting to do this I discovered the value of it almost immediately because I found errors in bridge numbers. I used the bridges as a reference point to measure out the milepegs and found pretty quickly that the bridge I had marked as 36 was actually a culvert and bridge 37 needed to be renumbered to 36. Then the real bridge 37 was at a point where I hadn't even noticed it up to now. The reason of course why it is so hard to identify the actual bridges is that small bridges and culverts (when the latter are built in a similar style, as some are) are often very similar in appearance and therefore it is hard to know which is which. Without the chainage charts you would have to try to spot them on the ground and some may well have their number labels missing in any case. Anyway fixing up where bridge 37 was meant an accurate measurement for the nearby 41 mile peg. The tricky question is whether to put the mile pegs in on the Taieri Gorge Railway as I can only do this back to Pukerangi which means about 12 pegs. But we may be able to estimate a few more from Dangerfield & Emerson's charts of bridge abutment or tunnel portal metrages. If I use those tables then perhaps I will have to put them in kilometres. Anyway putting those in is not a high priority at this stage, only for beyond Middlemarch to start with.

From that point of having fixed up some stuff and putting in all the milepegs starting from 40 at Middlemarch (which is a chain further westing compared to where I marked it as the end of line, being effectively in the Cardigan St road corridor) through towards Hyde. Why am I putting so much detail into the Otago Central compared to a lot of other rail corridors (live or dead)? Mainly because I want these maps to be useful to a lot of people, and they can potentially reach a much wider audience than the maps of the other routes. The OCRT is our biggest rail trail, and I hope to cycle it myself one day. Having all this detail in appeals to railfans like me who want to know what was happening at a particular place, and this is why I am working so hard on this route. Actually the work is at about the same level as what I have been doing where possible with other routes but it does have more of a significance for the rail trail because of its nature compared to a live rail route. The use of the modern Linz aerial photography is something I am just starting to do a lot of and I am picking up details on the ground because of it, since outside main centres GE resolution isn't that great. Of course the Alcam maps also show good detail and using them is what has convinced me of the merit of downloading the Linz stuff.

So a lot of today was fixing up bridges and milepegs. The bridges I got from Linz were misrepresented in a lot of cases (way longer than they actually were). Obviously Linz employs some lazy cartographers going by all the issues I have to fix up with their data to make my maps a high quality. I also found this recently with Main North Line aerials of the earthquake damage around Kaikoura. So it is not beyond reasonableness to use their aerials to fix up a lot of maps, but I doubt I can be bothered doing all of them this way, especially the North Island stuff.

Thursday, 19 January 2017

Otago Central Railway [3]: Middlemarch-Rock and Pillar

Today I started work on the rail trail route from Middlemarch up. So far I have completed about 13 km, which is about 10% of the rail trail route. Along the way there have been bridges to check and locate, and three major sites to draw on the map.

It is clear that the chainage charts are an incomplete record and many features have been added through checking details of photographs in addition to the charts. More research would be needed to have a fuller picture of each station but as my resources are limited the only other option I have is to place a disclaimer notice on each yard layout over which there is any doubt.

Tisdalls Ballast Pit

This was located between bridges 38 and 39, just south of Ngapuna station. D&E record that ballast was loaded for use as far back along the line as Pukerangi once the tracks were connected up. I have seen a reference to a Middlemarch ballast pit but it is not known if this is this location or a different one. D&E state it closed 1934. According to the chainage charts, the points were removed and reinstalled at Hyde, where a new ballast pit was opened the same year.

Ngapuna Station

Ngapuna is 7 km north of Middlemarch. It was closed to rail traffic in 1979. The small shelter shed was removed from the site after that date but has been reinstated for the rail trail sometime between 2011-12. It had a short loop with a backshunt at the westing end which followed a slight curve in the main line. 

Rock and Pillar Station

This location is another 7 km further on from Ngapuna and adjacent to the SH87 road crossing which probably gave it more convenient access. At this stage it appears that Rock and Pillar had more facilities provided than Ngapuna, perhaps because of the better access direct off the highway. Not shown on the chainage charts are the houses nearby, the water tanks that supplied them, and a building on the platform next to the shelter shed which appears in a D&E photograph. The station lasted until 1981. There is nothing at the site today but a carpark has been nearby formed for the rail trail.

Wednesday, 18 January 2017

Otago Central Railway [2]: Wingatui-Pukerangi

Well the main effort on the Otago Central has naturally started from Wingatui and worked up. Yesterday I spat out the first set of maps for trial and these are now up at the new nzrailmaps Flickr site ( Today's effort has been to work over them all again and fix numerous issues.

The main issues worked on today are:

  • Location of the junction between Kiwirail and TGR. Originally this was going to be 4 km but I am confident a number of sources put this as being 3.5 km now. But I had to move it about 100 metres further north on the maps.
  • Location of North Taieri crossing loop on the TGR. Which I put at 4 km. See next entry.
  • Location of North Taieri Tanks on the OCR. This I am tentatively marking as a little short of the School Road crossing, which is right at the bottom of the Salisbury bank. It is given by Emerson and Dangerfield as 4.36 km. On their website, Dunedin Railways amalgamate North Taieri and North Taieri Tanks into one location which they give as 4.4 km, but I am satisfied for a number of reasons that they are far enough apart to be marked separately.
  • Location of Wairongoa flag halt. Again this has to be marked as an uncertain location, but a photo in D&E makes it a bit easier to nut out.
  • Bridge numbering. A curve and gradient diagram confirms the location of Bridge 2 being near the start of the TGR. It looks like the three bridges just above Wairongoa would be something like 3, 4 and 4A, but the writing is a bit faded. There is another bridge 4 something just above Salisbury but again I can't read precisely what so with all these bridges I have had to leave the actual numbers off.
  • Position of Salisbury. Which has been confirmed from a photo as being on a curve, rather than where the hut is today which is on the straight just a little further round. Also a small bridge was put on the map just above the station as the C&G diagram shows there is one there, but I couldn't make out the number on the diagram.
  • Position of Taioma in 1928 which I decided would be closer to the road crossing aftter looking at the NZMS1 map. 
  • Position of Taioma in 1890 remeasured off the end of the Wingatui Viaduct. There turns out to be a very short straight between two curves and apparently the station was on that straight, although of course I have not been there to look, but according to an older edition of D&E the concrete edging can still be seen there.
  • Added the houses at Mt Allan
  • Marked positions for the station building and stockyards at Mt Allan based on a photo.
  • Confirmed two locations for Little Mt Allan, one of those is the former tanks on the easting side of the of the viaduct, the other is the location from the NZMS1 map, westing of the viaduct. There was apparently a passenger halt which may or may not be the same location as where freight was loaded. 
  • Check location of Christmas Creek by measuring off viaduct.
  • Try to determine where bridge 9A was at Hindon. This apparently went over the railway line somewhere near bridge 9 (the Taieri River bridge) and was used as part of a stock route that also went along the upstream side of the No.9 bridge between 1911 and 1966. This was all the access that was available to the other side of the Taieri River from Hindon until a road bridge was put in in 1966. This same bridge had to be demolished around 1993 when it was undermined by floods and that is when Bridge 9 was converted into road and rail use which it continues as today. Unfortunately the NZMS1 maps do not give any clue as to where bridge 9A was and the writing on the C&G diagram is too hard to make out. It's possible the bridge went over the cutting at the westing end of the tunnel but I can't be absolutely sure.
  • Confirm locations and bridge numbers for bridges 11-14. The map had marked bridge 13 as 12 and 12 as 11 while missing out the real 11. Some hard work of measurement found bridge 11 to be just next to bridge 12 so adding one more small bridge and renumbering the others was needed.
  • Locations were put in for Pay Office Creek and Machine Creek which were watering halts between bridges 12 and 13.
  • The location of Deep Stream was confirmed. Arthurs Knob was confirmed and renamed to Arthur's Knob.
  • A placemark was put on the map for the Notches so the location would be labelled clearly.
  • The phantom bridge between B15 and B16 was removed.
  • Both locations of Flat Stream were checked.
  • Quartz Reef Siding was looked at but not rechecked. Which means I didn't try remeasuring it from The Reefs which must be how I placed it first time in GE. Looking for details of the mine drew a blank.
  • The location of Barewood Mine was clarified.
  • Pukerangi station building was altered following an examination of a photo in D&E which showed there was an extension at one end (ladies waiting room) that has since been removed.
The above alterations brings the first part of the TGR maps up to the standard of the second part (Pukerangi-Middlemarch). The work on this latter section was done progressively over the past few days once I had obtained aerial photography of this latter section of the route, and in conjunction with the chainage charts, I was able to mark on all the bridges between Pukerangi and Middlemarch. Yesterday I put in all the sidings and facilities at Sutton and Middlemarch and also marked in the station building at Matarae station.

At Middlemarch I have also marked an end of rails. Unusually this is labelled as 40 miles 0 chains based on the chainage charts. I am not confident that I could give this in kilometres without being able to locate the 64 km peg first and since I don't know where that is I have stuck with 40 miles for now.

Well the next stage is to start working on the rail trail and to achieve the best outcome I am obtaining current LINZ aerial photography for the entire line. This is going to assist with the accurate location of various details along the trail. One benefit is in having the DOC conservation and science notes as well as the chainage charts to locate all of the bridges. So that work will be going along from here.

Project Update: Otago Central Railway [1]

Well over the break work first began on updating the MSL from Palmerston to Dunedin and after getting as far as Oamaru work on that has stopped for the time being to focus on the Otago Central Railway. This has been intensified over the last week looking at all areas of the line but naturally the actual map work has been undertaken starting from Wingatui.

At the same time I have been floating some ideas around for publication and it has fallen on me to look at ways of reducing the size of the published volumes down to something more realistic. To do this I have tested two formats that split an A4 page into something long and narrow. The two formats that have been experimented with have been to split the page horizontally into two pieces approximately 285 mm high and 100 mm wide; and to split the page vertically into three pieces of about 90 mm high and 200 mm wide. 

The latter format is the more appealing of the two as it fits more naturally with the shape of regular desktop computer screens, meaning the same format can be used for both. When the images are put into a Flickr album they can be swiped across very naturally and intuitively on a handheld device such as a phone. I tried the swiping thing before but there were issues where the map layout inconveniently had the track exiting the top or bottom of the image instead of the left or right side. To counter this and also make best use of the available space the second innovation is to rotate the maps. This required a rotating north arrow which is a capability that has recently been added in Qgis, so the map will have an arrow in the lower left corner that always shows where north is.

The third improvement is that again Qgis has been updated so that we can have a scale bar that is marked in metres even though the project coordinate reference system (EPSG:3857) uses degrees, minutes and seconds as its units. Seeing an actual distance marked on a map is better for two reasons: it is easier to work out distances with a scale bar marked in metres than it is by trying to do impossible calculations when you don't have anything that inherently tells you how to convert a ratio like 1:8000 into something meaningful, and if the original map is scaled to a different size for some reason, the scale bar distances will also scale proportionately to the new size.

So things have got a lot better than when I first experimented with producing printed volumes that used two A5 size maps that was not an efficient use of space and resulted in up to 150 pages for a volume. The maps that cover the Taieri Gorge Railway for example can be done in about 12-13 A4 pages if that was the format chosen and this is a lot better than what it first looked like. The entire OCR could be done in about 50 pages. Anyway we will have to see where that goes, at the moment I am just pushing ahead with the maps.

Here is a sample of a map with full footer (as will be put on Flickr). For a publication only the scale bar and north arrow would be included. The footer references the new Flickr site I have set up for the maps where the first full set of maps for the TGR can be seen. The example is Middlemarch station (more in the next post).

Tuesday, 10 January 2017

MSL Waitaki-Palmerston [1]

Well back onto another MSL series. Right now the MSL is not fully detailed between Christchurch and Dunedin. There is about a 60 km gap between Palmerston and Waitaki that is not fully detailed. I have decided to fully detail this section and probably after that, from Dunedin to Invercargill, while waiting for a response to my approach to Railfan Magazine over the Otago Central line project. Detailing this section of line will require adding the sidings and loops between those locations using the Linz aerial photography of the area. Once the photography has been downloaded and added to the map project adding the extra tracks is very straightforward.

So I have just picked out and downloaded the aerial photography for Oamaru to Palmerston, covering Pukeuri, Oamaru, Herbert, Hillgrove, Bushey and Palmerston. I expect it shouldn't take more than today to put in the loops and sidings in these places. Of note is that I have included in the Oamaru download enough coverage to take in the Ngapara and Tokarahi Branches. But I haven't bothered with the Kurow Branch or the Dunback / Makareao Branches.

Palmerston is an area I've already passed in detailing, but I have gone back to put in extra detail. Unfortunately the available coverage isn't detailed enough to see where the siding tracks are. Extra tracks were put in when Oceana Gold began mining at Reefton and railed their ore east and south to Palmerston for processing at the Macraes site inland. This traffic has ceased at the time of writing with the Globe Progress site at Reefton going into care and maintenance wind-down and Palmerston will go back to being sleepy hollow again. The loop at Bushey, the next station north, is where trains have been crossed and when the Taieri Gorge Railway were running locomotive hauled trains to Palmerston a few years ago it was necessary to run on to Bushey in order to run around the train for the return trip. With the extra sidings at Palmerston of late and using a railcar instead of locomotives and carriages it has not been as necessary to use Bushey for this operation. What will get put into Palmerston are the major buildings and the Bushey loop, and I will try to come up with a location for the former turntable (which went to Feilding about ten years ago) as it was still there in 1999 and I have photos of it from that time.

As I noted in my previous post, issues with the current master version of Qgis make for a complex set of steps to add new features to the map. All of the detail work has to be done on my PC, because VMs don't have enough resolution of the mouse clicks to be able to precisely place stuff like on the application running natively on the computer. But, creating new markers, lines or polys in 2.99 runs into problems when saving the new object because of a validation bug to do with certain fields that are present in almost every table of the database. So at that point I have had to install Qgis 2.14 onto my Windows PC and then use that, which is slower because of its low spec CPU and also because it loads the files over the network from the main PC.

This problem was fixed in a later build of the development master but we have a new problem, that there is a limit in the Linux edition on the number of layers which I discovered when I started adding lots of aerial photography images to trace over. So I can only load a certain number of images at a time in 2.99 running on Linux (there is no problem at all with Windows, where I can load 1000 layers easily, but performance is so much slower on that computer) and that has been my workaround to let me keep going with 2.99 on Xubuntu. A further problem has arisen due to issues with the Qt component library that Qgis is authored with and that is that in 2.99 there are rounding issues with the displayed distances on stations (which are stored in the shapefile's database table in kilometres to up to 4 decimal places) - the labelling is showing 42.70 as 42.69999999999999 km which is really annoying. To get around this therefore the actual production of map images has to be done in a VM running an older version of Qgis.