Salonika front maps , Datum and projection

[Stavros]

I am a new member from Greece ,

Hello to all the company , 

 

Does anyone know who was datum and projection used by the topographic teams to compile Salonika front maps ?

[Gardenerbill]

Hi Stavros,

Welcome to the forum, I have a trench map CD from the Salonika Campaign Society that contains information about the map creation, I will have a look later today and see if it answers your questions. 

[Gardenerbill]

Hi Stavros,

Datum and projection are not specifically mentioned in the information on British Military map making in the Balkans. If you PM me your email address I will send you a copy of the article from the CD, but before you can use the Personal Message function on the forum you need to make a few posts. A couple of replies to this topic should do the trick.

[Stavros]

Hi Gardenerbill ,

 

I have these cd from Salonika Campaign Society , they really do not mention anywhere about Datum and the cartographic projection of the maps, data absolutely necessary for someone who wants to determine with a very good accuracy, some point of interest on a map making through some special applications the georeferencing process.

 

Nevertheless, thank you very much for your time.

[Gully Ravine]

Stavros raised this with me shortly after purchasing the maps.  I have done quite a bit of reading / researching into this, without success.  Learned quite a bit about mapping - just not helpful to the specific question.

 

I would like to find some method of converting the grid references on WW1 maps (specifically those relating to the "Eastern Front") to modern latitude and longitude.  Starting with (for example) a reference of 117 / 190 as used on one of our maps on the DVD without any additional "leading codes" is not that helpful (so far).

 

Any help appreciated.

 

Keith (SCS Mem Sec)

Edited 23 October , 2020 by Gully Ravine

[WhiteStarLine]

Hi all,

 

I've been in discussions with Stavros since August and he sent me some material referenced in the above posts.  There are two approaches to this but first, a big apology to @Stavros.  I totally forgot that a few years back, when I was researching formulae for converting Western Front references to WGS84, my first visit was to the work of Mugnier[1], whose Grids and Datums columns give the history of each country's geodesy from the earliest times to the modern day (1980s).  I only remembered this useful reference series after commencing this post!

 

So the first approach is to find the spheroid of reference, the projection and the point of origin used by the WW1 military surveyors.  I've offered to assist Stavros but I won't be able to help actively until some of my Western Front map tiling has completed.  If someone wants a go, here is an approach that generally works.

 

These days, petroleum or remote sensing is the principle driver of mapping, but in our time of interest, the datum and projection was military-driven, so find what the surveyors used in the country of interest at the time of interest and you probably have an answer.

 

Determine Spheroid of Reference

 

Mugnier quotes:

 

   "In 1889, the Greek Army Geographical Service was formed, and classical triangulation commenced immediately. The agency name was later changed to the Hellenic Military Geographical  Service (HMGS). The initial starting point for the triangulation was the Old Athens Observatory where Fo = 37° 58' 20.1" North, Lo = 23° 42' 58.5" East of Greenwich, and was referenced to the Bessel 1841 ellipsoid where the semi-major axis a = 6,377,397 155 meters and the reciprocal of flattening 1/f = 299.1528128."

 

   "The Greek Military Grid was based on the Hatt Azimuthal Equidistant projection, a system originally presented on the sphere by Guillaumme Postel."

 

The use of the Bessel [2] spheroid of reference isn't far off the reference system used on the Western Front (Carte de France) but their projection (Bonne) was widely different.

 

Determine Map Projection

 

It is a reasonable assumption that the survey notes cited by Stavros relied on the extant projection with French astral fixing used for survey points.  With a bit of luck, this is a good starting approach to find their point of origin.  When converting the Western Front, I found spherical geodesy to be unusable but Snyder's ellipsoidal formulae [3] to be absolutely spot on over a tens of thousands of square kilometres (given a WW1 standard grid had a best case 25 square yard coverage before any margin of error).

 

Snyder's ellipsoidal formula for Micronesia was his implementation of the Hatt projection and would be a good starting point for distances below 800 km.

 

Determine Point of Origin

1. I would start off by finding the places cited by Wood [5].  Particularly Vardar Plain and the White Tower, Salonika.  Use this to determine a point of origin.
2. Record this origin in WGS94 latitude and longitude as the center of projection and origin.
3. Find a length between the origin and a known point in 1916 metres (x and y, measured from WW1 map in Bessel CRS) using the grid coordinates of the trench map.
4. For simplicity, use Snyder's inverse spherical formulae and the Bessel sphere.  Your results should look very roughly OK but nothing like the accuracy you want.
5. To see if you have really made it work, try the inverse ellipsoidal formula.  These take me well over an hour to setup correctly (noting carefully any arctan assumptions to save much grief!).
6. Do a few and you will spot a pattern - the linear translation you need to make it all align on a WGS84 grid.  Technically this is a Molodensky transformation, but I never bothered with a 7 parameter one over the distances we are looking at.  Factor this into each conversion.  If you are right, then you get an accurate conversion over considerable distances.

There is a much simpler method and I will leave that to the next post.

 

References

 

[1] Clifford Mugnier, The Hellenic Republic, part of Grids and Datums series, 2002 Louisiana State University, 

[2] F. Bessel The calculation of longitude and latitude from geodesic measurements, Konigsberg Observatory, originally published: October 1825

[3] John Snyder, Map Projections - a Working Manual 1987,  See p 210 for "Clarke's best formula" of 1971 and p 337 for a worked example.

[4] EPSG ESRI:54032 World Azimuthal Equidistant

[5] "The British Triangulation in Macedonia 1916 – 1918".  By Lt Col H Wood, R.E. 15 November 1918

[WhiteStarLine]

 

8 hours ago, Gully Ravine said:

some method of converting the grid references on WW1 maps (specifically those relating to the "Eastern Front") to modern latitude and longitude

 

The simplest method is to take a small section with known points and fit it over a modern map.  Then you have different base map layers (like Streets or Aerial Photos), Distance & Bearing and opacity sliders to help you.  Navigate with pan and zoom.  This example had a 150 year old tree and well that Stavros was interested in.  In the modern map it is very visible, close to the modern road.  You can see the entire image at GeoRef and while this site is currently aimed at Western Front maps, only a few modifications were needed for it to be just as effective for other theatres.

 

 

 

This approach is used by National Library of Scotland and some commercial products for the Western Front.  It requires no mathematics, just patience and enthusiasm.  I've just georeferenced and tiled 20 trench maps for Flanders, each of 8 km x 5 km.  If the copyright owners (SKS?) upload 1 or 2 trench maps I am happy to experiment by georefencing, tiling and returning the maps as tiled overlays suitable for hosting with modern online maps such as Google or Bing.

 

If they are not prepared to do this, an alternative appears below, with a caveat that there is a real learning curve.

 

Kavak (click to enlarge)

 

 

Alternative Approach

 

Note that this approach works well for France / Flanders where many towns, railways, church spires and rivers are in exactly the same place they were in 1916.

1. Download a suitable product.  I use the Open Source product QGIS  The free or paid version of MapTiler is highly regarded.
2. Respecting copyright, find a trench map or trench map extract.
3. Create a new project and add Open Street Map tiles.
4. Find the location (or insert the Open Street Map search plugin and find it).  Open Street Map is good with modern spelling but these may not relate to WW1 place name spelling.
5. If you have a complete trench map, trim the margins.  This makes it easy to follow roads and rivers so they don't 'disappear' under the map margin.
6. Use the Raster Georeferencer and open the trench map.
7. For an entire trench map, this might test your patience!  Find around 6 points common to each and click on the trench map location then the modern map.  Close to each corner is best.
8. Click the georeference wizard and don't overstretch the precision or you get some weird looking fits.
9. Zoom in, pan around and use the opacity slider to reveal modern and WW1 location.

 

Options

1. Optionally convert into what are called XYZ tiles, used in modern mapping products like Google and Bing.
2. Save these to your local hard drive.
3. Ask QGIS to add a Map Viewer
4. You are now ready for a field trip with a complete georeferenced map and an HTML page to run it.
5. You can even add Location Services so your table, phone or PC shows you where you are on the modern or trench map.

 

 

[Stavros]

Very useful the information you have written Bill ,
Also valuable, extremely simple and easy tool : http://georef.tmapper.com/

But the problem lies at this point: on some map sheets, there are no fixed points from the time the maps were written until today, because they have faded, changed dramatically, or even disappeared.
That is why I am doing research to find out these valuable elements ( Datum and projection ), but also the mathematical relations, which connect the Cartesian - orthogonal coordinates that all the sexes of the series have, with the geographical coordinates ( N , E ) so that in the first year this conversion is done first, and then the conversion datum and projection . and all this with the aim of maximum accuracy.

There are applications that with a few clicks, give you the ability to geo-report too many maps, with very good accuracy, and very little time (as long as one knows the Datum and the projection )
There is also the possibility for someone to create '' user datum ''  as long as you know the permutation parameters ( DX, DY, DZ , DA , DF )
Bill has already mentioned an application that does all this and I have heard a lot of good Cretans ( QGIS )
I use the application https://www.the-thorns.org.uk/mapping/ and I am quite satisfied.

Today, looking in my file, I noticed that I have a 20K map gender. by the Hellenic Geographical Service of the Army, which has been reproduced with the background of its respective sheet Armee d' Orient.
The important difference is that the Hellenic Geographical Service of the Army, while it has for the same area all the data as shown by its map Armee d' Orient the coordinates at the top of the map, are geographical and not rectangular, this to have I think that the Greek side had received all the necessary information that was written above the allied army, so that with the same background the old observatory of Athens has a starting point (I attach the relevant)

So within the next week, I thought of sending a request to the Hellenic Geographical Service of the Army, lest this valuable information be leaked, and we can go one step further.

 

Edited 24 October , 2020 by Stavros

[Stavros]

Unfortunately, the answer I received from the competent military service was negative, and this is because during World War II the Germans took all the relevant files and equipment, which were in the service.

[WhiteStarLine]

What a shame Stavros.  At least you tried and we can rule out that approach.

[Stavros]

Exactly , WhiteStarLine thanks ,

 

However, I do not stop trying to find the information I need at some point.

For anyone who will find the relevant data for the Western Front useful, I have found a very valuable archive which I attach.

 

 

reportonsurveyonwesternfrontmoderntext.pdf

[WhiteStarLine]

Hi Stavros,

 

Yes, that report is the holy grail for conversions in northern France / Flanders and it is the basis for Western Front mapping.  It gives the coordinate reference system, survey points, datum and how these were arrived at.  It is an essential read for anyone interested in WW1 mapping.

[Holger Kotthaus]

On 24/10/2020 at 16:18, Stavros said:

Also valuable, extremely simple and easy tool : http://georef.tmapper.com/

I use the application https://www.the-thorns.org.uk/mapping/ and I am quite satisfied.

 

Hello Stavros,

 

many thanks for this helpfull references!

 

Cheers Holger

[Gully Ravine]

I'm not sure if this has been included before.  Extracted from:

The British Triangulation in Macedonia 1916-1918

Co-ordinates of the points were computed on the rectangular system, the French value of the bearing of the side Two Peaks – Deve Kran (36° 24’ 31”) being used in the computations.  The White Tower, Salonika, was used as the initial station and the French co-ordinates for it (X 144,810m., Y 123,359m) were accepted.  [The French values were obtained from a system of triangulation executed by that nation emanating from a base measured in the Vardar Plain, with an astronomically determined azimuth.

Arbitrary values for the X and Y co-ordinates were given to one end of the base.  Subsequently the azimuth was found to be considerably in error, and to save re-computation this error was considered as if due to convergency and the co-ordinates of the point which, taken as an origin, would give this convergency were calculated.  This point (X 173,000m., Y 129,000m.) is therefore the true origin of the system].

The geographical co-ordinates have been converted to rectangular co-ordinates.  The latitude and longitude of the point of origin of the French system (X 173,000 metres, Y 129,000 metres) was obtained by taking the naval values for the geographical co-ordinates of the White Tower (latitude 40° 37’ 31”, longitude 22° 57’ 41”) as equivalent to the French rectangular co-ordinates (X 144,810, Y 123,359) of that place and computing the geographical co-ordinates of the French point of origin.  The rectangular co-ordinates of the point of origin and of the White Tower are taken from page 1 of the list of co-ordinates published by the French.

 

Can anyone explain what it actually says please?

Keith

[Stavros]

Having to deal with the problem of the correct translation of a language other than my own, what I have found "as essential" from this text, but also in general from all the relevant information contained in the 2 dvd (S.C.S.), I mean that:

a) The initial measurements and the trigonometric network, had significant deviations in several points, and for this reason the whole network was re-measured from the beginning, and the errors of the first measurements were shared proportionally, and based on the new measurements a correct trigonometric network (from the 2nd half of 1917 onwards)

b) The main problem for us today, is that there is not the necessary information that accompanies in each case the values of coordinates given (this was introduced worldwide a little later).
That is, if we enter the coordinates given for the White Tower of Thessaloniki (latitude 40 ° 37 '31 ", longitude 22 ° 57' 41") we will see a deviation Mainly East 1,110 meters,
This is because the "Datum" used by Google Maps today (it is wgs84) has very different parameters from that used by the Allies, "and nowhere does it say what it was, with what I have searched so far".
So .... the same coordinate values can have as many different positions in the real world as the different Ellipsoids, or Cartesian coordinate systems.

c) Somewhere on the 2nd dvd (5 Maps - their scale and distribution) there is a report that the vertical lines (meridians) do not correspond to the true North, but the problem is that then this information is not specific at all, because it does not states what is the difference (how many degrees?) of the meridians of the grid (North square) with the true North.

Either way, I personally will not stop working and looking for the information she needs.

[Gully Ravine]

Its certainly an interesting challenge.  Over the past few weeks, I have made thousands of calculations using various different formulae / estimates etc etc and still feel no nearer to solving this problem.  I wish to add some more information to the SCS DVD collection and can't achieve this until this specific data is known.  Like you, Stavros, I'm not prepared to let it go unresolved.

Keith

[MaureenE]

A short general account, available online 

"Macedonia", page 91 Records of the Survey of India, Volume 20. The War Record 1914-1920

https://archive.org/details/records-survey-india-vol.-20/page/91/mode/2up

 

 

[Stavros]

I consider any information to be valuable, but I do not think this report mentions "important technical information" either.

[Gully Ravine]

I hope the following is clear and helpful, but please note geodesy is not my area of expertise.

In trying to resolve this query, I have been given great assistance by two eminent Professors, both experts in this field.  My thanks go to Professor Cliff Mugnier (Head of Geodesy (Emeritus), Louisiana State University, Baton Rouge LA, USA) and especially Professor Gábor Timar (Department of Geophysics and Space Science, Institute of Earth Sciences, ELTE Eötvös Loránd University, Budapest, Hungary) who were extremely knowledgeable and generous in their time.  As a result, I am able to post a summary of their findings.

The White Tower, Salonika, was used as the initial station with the French co-ordinates for it (X 144,810m; Y 123,359m) obtained from a system of triangulation emanating from a base measured in the Vardar Plain, with an astronomically determined azimuth.

The Latitude and Longitude of the point of origin of the French system (X 173,000 metres; Y 129,000 metres) was obtained by taking the naval values for the geographical co-ordinates of The White Tower (Latitude 40° 37’ 31”; Longitude 22° 57’ 41”) as equivalent to the French rectangular coordinates (as above) of that location and computing the geographical co-ordinates of the French point of origin, as follows:

Geodetic Datum: ellipsoid: Bessel 1841; Molodensky-type shift parameters: dX=+923 m; dY=-809 m; dZ=+535 m

Map Projection: type Bessel, parameters: lat_origin: 40.62528; lon_origin: 22.96139; false easting: 144810; false northing: 123359.

It is estimated that the accuracy of these calculations is approximately 80 – 100 meters.

Keith Edmonds

Membership Secretary, Salonika Campaign Society

[Stavros]

A big thank you from me for the effort and the result.

Definitely something useful and will help a lot !!

Edited 27 January , 2023 by Stavros