CTS 91.1327/f: Cambridge Underground 1991 pp 28-31

Surveying Report Part II - Survey production


This is an attempt to summarize what I have learned about the art/science of survey production, from getting people to do it in the first place, through processing the data, to producing and distributing the finished product.

I have only performed this exercise twice so I don't claim to be a great expert, but I think it is worth recording the current state of my art (?) to give future CUCC surveyors something to work from. Having said that, I suspect that much of what I will say will very rapidly become out of date due to the extensive use of technology.

First, never forget that surveying is ultimately pretty pointless, and that most of your compatriots have very little interest in surveying for its own sake. These people can usefully be bullied and cajoled into going out there and getting you some data, but it is fair to say that an uninterested surveyor is not likely to produce outstanding work. Bearing this in mind you are likely to have to go and do lots of real surveying yourself. You will then notice that it is very hard to do well, and that the only reasons you understand your stuff any better than anybody else's are a) because you did it yourself, and b) because you learned what information you need from having to make half of it up the year before.

It is almost impossible to appreciate exactly what is required to make drawing up straightforward unless you have tried it yourself, so any opportunity to make your surveyors do some should be utilised (I have largely failed in this so far).

Right, that's enough sermonising.

Having got yourself some data what do you do with it? Encourage your surveyors to follow helpful practices when writing it up in the survey book, and then have a look over it. You may immediately find missing information - calibration, compass numbers, names, dates, locations etc. This sort of thing is often easy to discover so long as it is noticed quickly. Reference to the original notes and/or surveyors should clear up most things.

Next stick it in a computer. If you can do this during the expo it obviously helps a great deal with spotting errors and with locating likely connections. Collar your surveyor, stick a plan/elevation of their bit on the screen and ask "does it look at all like that?" Answers like "No" and "That's not my bit" should be dealt with by working out which bits are backwards/total bollocks so that this can be allowed for or corrected if feasible.

Current CUCC computer policy is dominated by the fact that our preferred surveying software only runs on a C/E/VGA PC, and the club doesn't own any of these. This situation is likely to change as we should shortly have some new software which should be persuadable to run on lots of things, and hopefully someone can eventually be persuaded to give/lend us some suitable kit (unfortunately the only offer I have managed to get so far is a Mac, which isn't terribly useful). In the meantime we will be using whatever people are prepared to take to Austria, or, if no-one does, it will be back to the club programmable calculator!

The software in question in Sean Kelly's Surveyor '88, written for the Queen Mary College Belize Expedition (as it unhelpfully tells you every time you run it). Make sure that you are using the improved '89 version which has had a major bug fixed and will actually do its sums right. (In the correct version SVY2POS has two copyright dates on the title screen - the second being 12 Nov 89). SURVEYOR.DOC gives instructions on how to use the program, which, although both basic and irritating in some respects, is essentially very good and does the job. So far as I know there is nothing better available, but Olly and I hope to address this as soon as possible.

Processing the data involves lots of comparing of cave plots with survey data and drawings to spot the obviously wrong bits. You can also use the errors thrown up by loop closures, but this often isn't actually much help as the closures tend be a little on the dodgy side anyway. This is primarily caused by 'expedition conditions' but as we are likely to be claiming Grade 5 it is worth encouraging the best practices possible, and to aim for errors to be under 2 to 3%. This is a rough guide, as the relationship between permitted error (within a grade) and traverse length isn't linear and so the percentage error could be more for long loops, and less for short ones. See the diagrams in Bryan Ellis's book.

Eventually you will have a plot you are happy with (although upon drawing up you are likely to discover more errors); and you will be very bored with plotting out 'absolutely the final survey', chopping up all the bits of paper, and sticking them all together. Unfortunately Surveyor '88 can only output onto an Epson compatible dot matrix printer (it might manage some slightly dissimilar interfaces but laser printers don't seem to like it at all, even when they claim to do Epson emulation. Until a method of getting the plots out on bigger bits of paper (eg. in a pen plotter) is created you are forced to a great deal of chopping up and sellotaping together of plots. This is very time-consuming and is absolutely desperate without access to a guillotine.

You may now come across some problems caused by adding bits to a pre-existing survey. Assuming that you haven't decided to change the view you should just be able to draw the new bits independently and graft them on to the old survey. Hopefully loop closures and discoveries of old errors will not have distorted the old stuff so much that this is impossible (as it did this year - aligning Adrian's put Yapate 15 to 20m out, for example).

The plan was to try and have something ready for the BCRA conference so I drew each new bit onto its own centreline (as this could be done in a tent whilst still on holiday), then traced each bit, reduced them onto acetate, and stuck the results onto an A4 acetate copy of the '89 survey. This sort of worked, but produced a fairly unhealthy-looking result as photocopying more than two layers of acetate produces lots of greyness. It is also a very expensive procedure as to get from the original 1:500 to A4 required a reduction to 22%. Normal copiers can only reduce to 64% (requiring four iterations), and plan copiers can manage 47% (so only requiring two iterations but they are much more expensive). To do all this for nine new bits of cave in both plan and elevation takes forever, requires the services of about six copy shops and is definitely not worth the effort or cost. Unfortunately I only realised this whilst half way through this epic process and decided that having already invested a fair amount of time, effort, money, and petrol, I might as well finish and have something to show for it.

When drawing the little bits it is helpful to just plot the bit you want to draw, but don't forget that you must arrange things so that whilst it is plotted on its own, it is still calculated as part of the total net otherwise it may 'unspring' significantly giving you an incorrect centre-line to draw on.

Having discovered that the old and new surveys didn't match properly and already having traced each independent bit I decided that the best way to do the final drawing was to put the Permatrace sheet on top of the final plot and then sandwich each transparent bit of cave between them. This allowed alignment with the plot underneath (as both top layers were transparent). The old master was treated in exactly the same way but each section between loop junctions was aligned separately. This worked very well, the only disadvantage being that by the time a bit of cave gets to the master it has been copied twice (or four times if you count the original as the one done in the cave).

If you are doing a new survey then the whole thing can be drawn on the final plot and then traced to make the master. The only disadvantage of this is that you have to work with a great big bit of paper lying around for weeks - OK if you have the space.

Obviously other combinations of wholes, sections and tracing could be employed - choose according to circumstance, remembering that tracing is extremely quick and easy in comparison to drawing originals (except for all the bloody rocks).

I have rather glossed over the bit which is definitely mostly art - drawing round a centreline so that the result looks something like the cave. There are actually a number of (sometimes conflicting) considerations here. Do you want it to be clear which way to travel or to have lots of realistic detail? This really depends on who you think your audience is and your own preference. I have aimed for authenticity at the expense of simplicity and make no claim that it is at all easy to follow!

A quick note on materials is probably in order here. Use a propelling pencil for the first drawings, as it has constant width and a rubber on the other end. The master needs to be drawn in pen so that it photocopies. Thick lines in 0.35mm, thin lines in 0.18mm. We tried 0.13mm in '89 and things tended to disappear on reproduction. If drawing at a very small scale you will find that 0.35 is just too thick to be sensible and everything will have to be 0.18. The club has Rotring isograph pens in the above sizes which have been very reliable and have not clogged up even when left from one year to the next. The ink these use is erasable which is obviously incredibly useful - but note that it is much easier to rub off just after it has gone down than a couple of days later. After being in place for a long time it will never rub off perfectly.

I recommend proper drawing office plastic film (eg. Permatrace) for masters as it is 'dimensionally stable', tough, waterproof, more transparent than tracing paper, and photocopies better than paper. The disadvantage is the cost - nearly £4 per A0 sheet! If using this then use the special film rubbers designed to complement it. Also required are a scale ruler (if you can't easily do it in your head) and a drawing board (the club doesn't own one of these yet).

So, you now have a complete survey drawn up - practically finished! Wrong. You may be halfway through if you're lucky. Doing the cross sections and the lettering is unbelievably time-consuming. Until someone can work out a better way you are restricted to getting all the names printed and then cutting each one out and sticking it on by hand. At least it's better than doing each letter individually with Letraset.

A bit more detail on this process. Use a sans-serif font which is nice and solid. Helvetica seems to be the most suitable of the commonly available ones. This is easy to do on a Mac, but requires some quite flash software on a PC as they have only just noticed that there are fonts other than Courier in PC-land. Getting all the names, and a suitable selection of question marks, pitches, climbs, vdlbs, too tights, and cross section labels out on someone's laserprinter should thus be quite straightforward.

To align them when sticking them on use sheets of graph paper carefully aligned and stuck onto the back of the master to give a grid. Be careful to align the text rather than the edges of the bit of paper that contains it. To actually do the sticking use the amazing Scotch Magic Tape. This stuff is totally photocopier transparent, can be written on and is the right stickiness so that you can peel it off again if you get it wrong, without tearing anything.

The tedium of adding text in this way is perhaps a good reason for attempting to scan the image into a computer so that a drawing package can be used to add the text. If anyone has access to a scanner then perhaps this could be tried next year.

The other 'little' thing remaining to do is the cross-sections (on the plan at least). First you have to pore over the survey book again and decide which ones you want, weighing various factors like how representative they are, which ones can be sensibly fitted onto the plan, and how much space/time/enthusiasm you have. Once you have chosen them you must mark them all, think of a numbering system, draw them all, and number them all (in two places, obviously). This does, of course take forever too, although it is a task suited to distribution if you have several helpers, as each can do their own bit and they can all be stuck on to the master later.

Finally, you must get your masterpiece photocopied and reduced to both the size people can stick on their walls, and to something that will go in the Journal, and Caves and Caving/Descent. This proved to be extremely hard in Cambridge, using the Xerox place next to Sainsbury's for the primary reduction and a much cheaper copy place on the industrial estate next to Tesco on the A45/A10 junction north of Cambridge. Having a slightly better quality master this year may have helped but I think it was the copy shop which just got it right first time. In Cambridge we had four visits to Xerox with several tries each time and lots of Tippex in between.

So there you are - piece of piss, and it only takes about 200 hours. If you'll just form an orderly queue of volunteers for next year....

P.S. if anyone thinks I am being dim doing things as outlined above, please tell me so. Obviously anything that improves either the quality or the efficiency is welcome.