, , , , , , , ,

So much has been going on lately, there hasn’t been much time to post on here! We’ve been trying out a petrol stove to take with us, sorting out the first aid kit, trying to remember some Russian phrases (a thankless task if ever there was one), buying/borrowing replacement equipment that we need, booking off-road transport from Murgab to the roadhead… Also a friend has been translating reports for us from the internet of recent (and not-so-recent) Russian/Ukrainian expeditions to the area (thankyou Zhelyo), we have been leaving instructions and useful contact details for our helpers back in the UK, trying to cut down on the weight we carry in to basecamp…. you get the idea. Also when time permits (i.e. not very often for me given my work situation) going running to keep fit for the expedition.

But I thought I’d blog about one of the more innovative things I’ve done this time around, improving our use of maps and GPS/GLONASS. Many expeditions to central Asia make do with google maps, but actually there are much better maps available. The quality of the surveying on the Soviet military maps is as good (in my past experience) as the British Ordnance Survey maps, and the map information is presented in a similar way with major and minor contours marked, crags marked, glaciers and permanent snowfields marked etc. Also the Soviet military maps are very beautiful maps, if you like that kind of thing… You can download small sections of the Soviet military maps in colour from www.topomapper.com and then stitch them together. Fortunately for us, when I became interested in the Muzkol mountains in Tajikistan 2 years ago there was a website available where you could download entire sheets of the map in one go and save yourself the time-consuming task of stitching all the small segments together. Alas, that website (www.poehali.org) is now part of www.mapstor.com where you have to pay for your maps.

Anyway, I digress. There is one problem with the Soviet military maps, they use a datum (i.e. grid system) that isn’t available with our GPS (or, if it is, it has a very different name on the GPS list of datums). So you cannot turn on your GPS, get a grid ref off it, then use that to place your location on the Soviet map. To solve this problem, you need to overlay the Soviet map datum (Pulkovo 1942 Krasovskii spheroid if you’re interested to know) with a datum that the GPS can use (e.g. WGS84 datum as used on google earth). In principle, the data you need to convert directly from one datum to the other is known (http://www.eye4software.com/resources/datum/4179/ ) but wading through all the maths to do the conversion “properly”, and then working out on which pixels to begin and end your grid lines would be quite time consuming.

Over the weekend I worked out and implemented a much quicker way to do it. My method utilises some knowledge from my other life as a physicist – knowledge of a free scientific software called Datathief. Datathief is a piece of software for reading numerical data values off a graph. So if as a scientist you see a graph in a publication and want to read off the data points to input into a theoretical model, or to compare to your own data, Datathief allows you to do this. You simply save the graph in question as a bitmap/gif/jpeg, and load it into Datathief. You calibrate Datathief by moving calibration pointers to three different locations on the graph, and inputting the x and y co-ordinates of these locations. Usually you would choose the origin of the axes and one point on each axis, but you can use any 3 points that do not lie on the same straight line. Then, you take additional pointers and move them over each datapoint you want the value of. Using the calibration, Datathief can give you the x and y values for each datapoint.

How does this help you work out where on earth you are during a snowstorm in the Pamirs? Well, what Datathief does mathematically can also be used to draw a new grid system on a map. Over the weekend, I loaded the electronic copy of the Soviet military map into Datathief and drew on gridlines for the WGS84 datum that our GPS uses. It is possible for Datathief to tell you where to draw the grid lines for any map datum you want, as long as you can calibrate it first by placing the 3 calibration pointers at locations with known co-ordinates in that datum. This is, of course, assuming that the datum is a simple square or rectangular grid, a reasonable approximation as long as you are covering a sufficiently small proportion of the earth’s surface for the earth’s curvature to be negligible.

Placing the 3 calibration pointers is where google earth comes in. Google earth gives you the location of any placemarks you put in, using the WGS84 datum. So you can calibrate Datathief by putting google earth placemarks at 3 points that are easily identifiable on both google earth and the Soviet map (e.g. summits). You then input the WGS84 co-ordinates of the placemarks into Datathief and you have your calibration. Of course, if your calibration is not accurate, then your grid lines will all be wrong so it is worth making the effort to ensure your calibration is as accurate as possible. You can do some or all of the following things to achieve this:

  1. Choose 3 points that are a long way apart on the map (preferably near 3 corners).
  2. Choose your feature carefully so there is no doubt as to its location – don’t choose a mountain with closely spaced twin peaks for instance, or an indistinct peak partway along a ridge.
  3. Google earth reproduces some features better than others. In the case of summits, you can get some idea of how good a job google earth has done by comparing the spot height on google earth to the spot height on the Soviet military map (in my experience, the correct spot height). I guess an alternative would be to compare the simulated image on google earth to real photos if any are available, and see how closely they match up.

Now, you can move the data pointers around the map (while leaving the calibration pointers where they are) and Datathief will give you the co-ordinates of the data pointers in WGS84 datum.


Soviet military map loaded in Datathief with calibration marker and data marker shown

So to draw a grid line (say that for 75 degrees 5 minutes) you simply move your data pointers to opposites sides of the map where they are both reading 75 degrees 5 minutes on the relevant axis, then open a copy of your map up in paint and draw a new gridline that connects the locations of these 2 datapoints. Repeat until the map is covered in all required gridlines. In practice, there are 2 factors that make this much quicker and easier than you might at first think:

  1. Your map will probably only cover a fraction of a degree of latitude and longitude, so you need to deal only in minutes and do not need to worry about converting minutes into fractions of a degree. You do, however, need to set google earth to give you lat. and long. In decimal fractions of a minute rather than minutes and seconds, or do the conversion yourself.
  2. You do not need to go through this for every single grid line. You can do it for a few grid lines and interpolate to get the starting and finishing pixels for the rest of the grid lines. The fact that you have only entered 3 calibration points means it is not possible to draw anything other than a simple square or rectangular grid in any case.

Once you have finished, you can check that you have got it right by looking at the google earth co-ordinates of other features and checking that these correspond to the co-ordinates of the feature according to your map grid.  The summit co-ordinates predicted by my grid lines on the Soviet map of the Muzkol mountains are within 50m of those predicted by google earth so this method does work.  I have just got the finished map with both grids printed to take to Tajikistan:


Finished map with original (Pulkovo 1942, black) and GPS-compatible (WGS84, red) datums marked.

Intriguingly, in addition to the grid on the Soviet maps (Pulkovo 1942 datum), there are markers at the sides of the maps corresponding to some other datums. One of these sets of markers lies tantalisingly close to the grid lines I have drawn, almost close enough to be from the WGS 84 datum…. but definitely not quite. There are literally hundreds of different datums in existence, so I guess it should come as no surprise that there are many that are quite similar…

I guess the final thing to note is the accuracy of this method is limited by the accuracy of the calibration, i.e. the accuracy of google earth. The Muzkol mountains seem to be surveyed very accurately by google earth (evidence for this is summit spot heights corresponding closely to those on Soviet map, and simulated images corresponding closely to real photos of the range), but if you had a range of mountains covered more poorly by google earth (e.g. the Torugart-Too mountains on the Kyrgyzstan-China border, where I climbed in 2010) it could be more tricky to get an accurate calibration….

John Proctor