Transdem - Georeferencing Poehali/Mapstor (Russian) Maps

[Vern]

I can't remember whether this was discussed on UK Trainz in the dedicated Transdem support forum, but if we did I've forgotten the answer.

As some will be aware, the Mapstor site offers a download service for a whole host of topographic maps of the world, annotated in Russian but easily understandable and preferable in Surveyor to using a Google or Open Street Map which often looks stretched and colours spread out very hard to follow.

The problem I'm finding with the Poeheli Maps however, is if you georeference according to the Lat/Lon in the four corners the purple grid does not follow the grid lines on the map and in Surveyor the detail can be "off" the DEM by some considerable amount.

I'm just wondering if there is some method to improve on the georeference accuracy. Obviously for populated areas or where there's lots of roads, alternate Lat/lon values can be found by clicking on the same location in Google Maps and selecting "What's Here?" then entering the value in Transdem. However for wilderness areas or those parts which Google Maps gives scant coverage (parts of the former Yugoslavia) this is not an option.

The Poehali gif files also have an option of downloading a .map data file with them, would this be readable or of any use to Transdem?

 

[geophil]

Vern,

Russian maps are usually Pulkovo/Krasovski S42/40. This applies to the geographic coordinates (lat/long) as well.

 

[Vern]

Thanks Roland, I'll try setting the view mode to that.

 

[Vern]

That seems to be causing a few issues.

Working on map sheet I33 129. Enter first set of co-ordinates 16 deg E 44deg 20min N okay. Second set is 16 30 44 20 but the value in the boxes self amends to 3 16 23.81E and 45 53 13.61.

Happens with Lat/Lon Secondary Grid checked or unchecked. Presumably I'm doing something wrong here?

 

[geophil]

Works fine for me. If using lat/long, the "secondary grid" box must be checked.

Gauss/Krueger coordinates will give slightly better results. Both the eastings and the northings have 7 digits (or even 8 for the eastings in the far east). A grid line is always marked with at least two digits, which are the thousands. You also need the leading digits which appear in small print less frequently on the sheet. A grid line marked "08" may translate to 46-08-000 (enter in TransDEM w/o the dashes). As this is classic Gauss/Krueger notation, the zone is part of the easting, the leading digit for Europe. Otherwise the easting would be 6 digits, as with UTM. So 4608000 means 608000 in zone 4.

 

[Vern]

Thanks Roland, I will have to try it again, though not for a few days as I'm currently working nights and the old brain gets a bit fatigued for problem solving!

 

[Vern]

Just a quick update - I still could not get the Poehali sheet to georeference using the Pulkovo system. I tried I33 130 and the same problem occurred on that sheet too. The same error persisted using Transdem 2.1. Workaround was to use the raster maps from Open Street which (amazingly) actually show the course taken by the old Balkan narrow gauge routes. Down side is these aren't so detailed and trying to pull all the maps off for a 150km route in one go gave a "follow up" error, so that section will need to be done separately.

As an aside, I presume there are no known web mapping services for the former Yugoslavian countries?

 

[geophil]

I checked with one of those 100k maps again. As these maps are scanned paper maps, there tends to be a bit of distortion, so 3+1 point georeferencing cannot always achieve an optimal matching. Furthermore, accuracy of the maps is somewhat limited, most are very good, a few are a bit more imaginary.

As I found out some of these sheets lack Gauss/Krueger grid coordinates, you only have lat/long. This has the disadvantage that do do not know the G/K zone which makes geo-referencing less precise. And, furthermore, the purple grid may not match.

We also cannot be absolutely sure about S42/40. As this is Yugoslavia, the reference system may be one the later derivations for Eastern Europe.

Have you tried the 1:200k sheets for the same region?

 

[Vern]

I checked with one of those 100k maps again. As these maps are scanned paper maps, there tends to be a bit of distortion, so 3+1 point georeferencing cannot always achieve an optimal matching. Furthermore, accuracy of the maps is somewhat limited, most are very good, a few are a bit more imaginary.

As I found out some of these sheets lack Gauss/Krueger grid coordinates, you only have lat/long. This has the disadvantage that do do not know the G/K zone which makes geo-referencing less precise. And, furthermore, the purple grid may not match.

We also cannot be absolutely sure about S42/40. As this is Yugoslavia, the reference system may be one the later derivations for Eastern Europe.

Have you tried the 1:200k sheets for the same region?

I haven't tried those but I did attempt to reference one of the US military 1:250k topo maps and that didn't go in precisely either - georeferenced okay, but the data off by several 100 metres - rivers on the side of hills etc. So it seems at the moment, Open Street Map is the best bet for Eastern European routes.