The Palestine Grid: configuring non-standard coordinate systems in MapInfo Professional

most of the following is taken from the 2007 paper I wrote on a GIS of Palestinian population movement during the War of Independence 1948

This tutorial will introduce 1942 Survey of Palestine maps and show you how to set up MapInfo Professional to support the Palestine Grid coordinate system. The instructions apply to MapInfo 8.0 through to MapInfo 10.0 and will also help if you need to configure other ‘obscure’ coordinate systems. I will be assuming basic knowledge of cartographic concepts. 🙂

The Survey of Palestine

First a little historic background: Britain started surveying Palestine and the Levant in the mid-19th century through the Palestine Exploration Fund. Here, two young, aspiring and later successful military figures also received their early assignments, H.H. Kitchener (assisting C.R. Conder in 1874) and T.E. Lawrence (1913-14 survey of Gaza and Sinai).

During and after the First World War, the British need for more detailed maps of Palestine increased. The northern region of Galilee had already been mapped together with the Lebanese part of Syria as part of the French Protectorate – unfortunately, based on the ‘difficult’ grid system of a French army Lambert projection.

Palestine came under British Mandate 1917 – 1948, with the Survey of Palestine department established in 1918. There are some photos of mapping expeditions, levelling instruments, theodolites – and a map of the 1946 triangulation system here.

My Palestine GIS was based on historic 1:100,000 maps from the 1942 Survey of Palestine, which showed Arab villages – albeit with labels in English transcription rather than in Arabic. Jewish settlements were often marked only with circles along with the proposed name in Hebrew, as they were often still at their planning stage. You can often find bounded areas of potential Jewish development:

excerpt of Metullah sheet, 1942 Survey of Palestine

section of Metullah sheet, showing bounded areas, circles and Hebrew labels, as well as transcribed Arabic village names (Survey of Palestine 1942)

The Palestine Grid in MapInfo

The Palestine Grid is a non-standard coordinate system developed by John W. Hagar, who first used the Palestine datum at ‘station no.2’ in 1928. As we will see later, the coordinates of this location will be inserted into the configuration file. Another system, the Palestine Belt, only differs from the Palestine Grid by its use of the Gauss-Krüger Transverse Mercator projection instead of a Cassini, as Dan Savage shows in his table of geographic coordinate systems. Also worth mentioning is that the United Nations (UN) produced smaller-scale 1:150,000 maps of Israel in the 1970s, using the military Universal Transverse Mercator (UTM) grid. Here, Israel is located in zone 36 of this reference system. As my GIS was going to be about 1948, I was quite happy to use these UN maps as a guide to locate Jewish settlements not yet established in 1942.

Cliff Mugnier (2000) wrote an excellent article (pdf) in Photogrammetric Engineering & Remote Sensing about the technical details of the Palestine Grid and related Middle Eastern projections and coordinate systems. If you’re after a more historic perspective, The Survey of Palestine under the British Mandate 1920-1948 by Dov Gavish (2005) is available on Google Books as a limited preview.

After scanning six paper maps from the 1942 Survey of Palestine at 300dpi resolution, I found a suitable geometry dataset of Asia to build the GIS on DIVA-GIS. This showed not only the Golan Heights but also the zone of non-engagement east of the mountains, supplemented with a dataset of major and minor towns from the MapInfo WFS Gazetteer. Because the ‘Palestine Grid’ is a non-standard projection, I needed to create a new algorithm in MapInfo’s mapinfow.prj file before I was able to register the scans. For Cassini-Soldner projections, MapInfo requires the following parameters:

parameters required for Cassini-Soldner based projections

parameters required for Cassini-Soldner based projections

Once you have gathered these together – for instance from Cliff Mugnier’s article – open the configuration file to insert the following MapInfo-specific codes, found in the MapInfo user guide:

30 (projection type) – MapInfo’s code for the Cassini-Soldner projection type
60 (datum) – To set up the datum, I was required to choose one from a list in the MapInfo manual. ‘Station 2’ was not listed, but with a known ellipsoid of Clarke 1880 I decided on Nahrwan. This covers the Arab penninsula, is the closest datum to Jerusalem and thus most likely to work.
7 (units) – MapInfo code for ‘metres’ (because we officially use the metric system in the UK)

The Lat/Long and Easting/Northing values are taken from Savage, Mugnier and – at least back in 2007 – they were also in the MapServer documentation:

31.180627 (origin, Longitude)
34.314202 (origin, Latitude)
170251.555 (False Easting)
126867.909 (False Northing)

This is what your mapinfow.prj should now look like:

excerpt from the Mapinfo projection configuration file

mapinfow.prj set up to support the Palestine Grid

And last but not least, a few quirks of this coordinate system in relation to MapInfo raster image registration. After registering the first scan to the geometry and gazetteer, it became apparent that the vector geometry was highly distorted. A little of this can be expected, after all the Palestine Grid is a coordinate system especially for Palestine and not suitable for the whole of Asia. What the Ordnance Survey has to say about Cassini projections, is that they tend to skew in a north-south direction and thus works best with countries that have elongated shapes. This can also happen when you have all your registration control points in a straight line – due to a lack of major and minor towns in the gazetteer file. I found no landmark dataset for the Middle East, the set of Geoname Features by the National Geospatial Intelligence Agency was too inaccurate and the elevation point dataset not detailed enough. I finally fell back on my trusted 1976 UN map of Southern Lebanon and Vicinity, scale of 1: 150,000. Allowing for registration errors of up to 40 pixels for the 1942 maps, here the error mostly lay between 10 and 20 pixels.

And there it is, the finished product:

MapInfo result of registering map scans, using two different projections

making both 1942 and 1976 maps fit exactly with the Palestine Grid projected vector geometry!

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s