There are many opportunities for caving organisations to use computing power to reduce the workload of administration - by printing address labels, wordprocessing articles for journals, or keeping up-to-date lists of books and journals in its library. However, all these applications are common to many sport or social clubs and the purpose of this article is to delve into some applications of computing which are unique to caving.

The earliest use of computing power was to reduce cave survey data to coordinate form before drawing up a survey. The reasons for this are simple: all but the simplest surveys need to be reduced to coordinate form, and the job is tedious and error-prone when done by hand, or even with a calculator. Thus, a surveyor with access to computer facilities naturally uses them to simplify calculation. This would often take the form of a 'foreigner' on a company or university computer - this is certainly the way I started.

The ready availablility of cheap computers means almost any caver can reduce surveys this way, but also provides opportunities for experiment, and the chance to devise more innovative uses of computing power without prohibitive costs. Some of these novel ideas extend the surveying theme, enabling larger volumes of information to be processed, perhaps even within the caves themselves.

The Notts Pot survey published with this journal was reduced from a mass of notes to an accurate centre line by a program which has been used on a BBC micro for about three years (Waddington, 1984). The input data for the program matches that for a mainframe program used for over four years previously. This means that the data can be used to give access to more expensive facilities that the home microcomputer user cannot afford. With the old program, I had to return home, punch the data onto cards, run the program and get results perhaps several days after a trip or expedition. Now, the micro can be used on Sunday night immediately after a weekend surveying: the results are in my hand in a couple of hours. On a trip to Europe, the machine may go with the expedition and be used 'in the field' to plot results and spot areas of interest for further exploration. The beeb was used in Austria in 1984 (Waddington, 1985): all the previous data were available, and new surveys could be compared with known cave, often on the day of exploration. Only the Beeb, disc drive, and monitor were needed, which today could cost as little as five hundred pounds all in. For an expedition, a printer was unnecessary, and hardcopy was only produced on our return.

The survey package used is known as SU-BBC, and works well for surveys of up to 1500 stations (in Austria, the total is about 6 miles). It takes about 20 minutes to process 1000 stations, and the resulting coordinates are kept on disc, to be inspected or printed at the surveyor's leisure. The whole survey, or any part of it, can also be shown on the screen from any angle. The package is available from the author, to run on a BBC micro or Master with discs. If anyone is interested in a copy, please contact me via Greenclose.

Beyond 'traditional' data reduction, the rapidly falling cost of graphical input means that a goal of surveys drawn wholly by machine is now within reach. Objections to this idea have been raised in the past because far too many extra measurements would be needed. This is true for a direct approach, such as used by the Swiss Toporobot program (Heller, 1980), but a little thought shows that the objection is really quite fallacious.

A surveyor can draw a good survey by hand from sketches in his notebook, so there must be enough data for a machine do so. The problem is getting the sketches into the machine. Digitising tablets have become much cheaper in the last five years: for example, there is now available a quite robust A4-sized device with better than 0.25 mm resolution for under sixty pounds. Once the sketches have been digitised, they may be cleaned up by a 'picture processor' by which the surveyor can move details in much the same way that he might do several drafts of a drawing. Once captured in this way, the surveyor need never draw the passages by hand again, and can be sure that the passage outline will not be degraded by repeated tracing. The survey can be transformed to fit corrections by loop closure or accurate radiolocation, and plotted output used either 'as is' or as a final draft for inking by hand. While this method is still in an experimental stage, an American program was producing good results applied to Mexican maze caves in 1980 (McKenzie, 1980).

So far, we have a tool to convert normal survey data into a drawing, but how do we get this information ? Computers are now so small that the idealist imagines using a small box with inertial navigation system and a powerful computer on exploring trips. Back home, connected to his home computer, out comes a beautifully drawn survey of the new cave. How far are we from this utopian idea ?

In fact, of course, we are a long way away: while computers get smaller and cheaper, the sort of precision measuring devices needed to carry out a survey do not. Experimental devices have nibbled at the edges of the problem, however. For example, a device in the USA does away with measurements to walls and roof at survey stations. Instead one presses a button on the box which activates an ultrasonic range finder as used in polaroid autofocus cameras (Breisch & Maxfield, 1981). This records the distance to floor, ceiling and walls on a cassette tape to which one can add verbal notes if needed.

A remarkable machine seen at the BCRA cave surveying symposium may come to the aid of cave divers, whose survey methods have inevitably been less accurate than those of their 'dry' counterparts. Diving gear being magnetic, all compass bearings are subject to potentially large errors. Distances must be measured by diving line tagged at fixed intervals. The surveyor uses his line belays as survey stations, but these do not usually coincide with the tagged positions, so distances must be estimated between tags. Depth gauge readings give fairly good data vertically. Nick Bennet's surveying tool lives in a long, neutrally buoyant tube held at arms length from the diver. It is away from magnetic diving gear, and also easy to align with the diving line. The diver has merely to press a button in a sealed boot at the end, and the device records its position. Unfortunately, the diver must still estimate distance by diving line, or survey between belays using a fibron tape. The latter approach can produce loop closures as good as a 'dry' grade five survey, even with a normal diving compass (Cordingley, 1986).

The machine, not surprisingly, contains a computer; in this case a microprocessor linked to purpose-built peripherals. A solid-state pressure transducer measures the depth below water level, while a biaxial magnetometer is able to detect the direction of the earth's magnetic field through itself, and hence, by calibration, the direction it points in. This could be made much more accurate than a diver's compass. Finally, ultrasonic transducers, similar to the polaroid rangefinders mentioned above, but for use underwater, measure the distances to floor, ceiling and walls, even if the diver cannot see them ! While still bulky, and definitely in an experimental stage, such devices will surely improve underwater surveys in years to come.

Aside from obvious numerical applications like surveying, there are various types of scientific study in caves which involve repetitive data-gathering tasks. Computers are ideally suited for this mundane sort of work, and in many cases, the quality of data can be improved by incorporating intelligence into the measuring apparatus.

A technique that seemed to have immense promise when first studied has remained the preserve of dedicated professional researchers owing to the very labour-intensive nature of the measurements needed. Flood pulse analysis (Ashton, 1966) has potential to deduce the nature of caves whose exploration is inhibited by long, deep or difficult sumps. However, the attenuation of small artificially generated pulses means that observations of water level at resurgence sites must be very accurate, very frequent, and very prolonged to be sure of detecting the arrival time. There are few volunteers to record the level of a resurgence to the nearest millimetre, every few seconds, for several hours ! With a micro to take the measurements, the researcher has an assistant who will sit for hours recording water level, temperature, pH, conductivity etc. etc. every second or so for as long as it has power supplied. Since a micro can make decisions, it can be made to ignore unchanging data, while recording immense detail during rapid changes, as when a pulse arrives. It is very simple for a micro to tell the time, so we could monitor a resurgence for changes of only seconds duration, while the gear is left for perhaps a week, during which natural storms may create pulses whose patterns can be analysed at leisure - almost certainly also by computer (Wilcock, 1968).

Temperature, pH and conductivity are easily measured with analogue to digital (A to D) converters, readily available to interface with a cheap micro. Water level can be measured in various ways, depending on the size of level change expected, and the precision needed. Pressure transducers, though perhaps a little expensive, can measure depth over a large range, but may not be precise enough. A simple version of the level used to record depth in reservoirs may, however, be fairly easily constructed.

Two metres (or more) of plastic gas main, or similar fairly large bore tube, sits vertically in water. Holes at the base allow the internal level to change in sympathy with the pool, while the tube protects a float/weight combination from disturbance by waves and ripples. The float drives a cord round a pulley connected to shaft position encoder. For rough measurements, this could be a multi-turn potentiometer connected to a reference voltage and an 8-bit A to D converter. Careful choice of pulley diameter could give 4mm precision over a one metre range, or better precision over a smaller range, as needed. This approach is relatively simple and cheap, and fairly easy to calibrate. Unfortunately, to detect small pulses at a site like Hurtle Pot, we need to measure the level to a millimetre or better at low stage, but still cope with a range of perhaps ten metres altogether. In this case, the device on the pulley should be an optical position encoder, connected to a parallel input port. The micro would count whole rotations and the output from the encoder could give very high precision.

Gear like this built from low-power components could run for a week on a motorbike battery, and to avoid data-loss if the batteries do go flat, would dump data to a cassette recorder at intervals. The latter would be switched off by a normally-open reed switch when not in use, and so would form a very small power drain. No doubt many readers can think of improvements to the device outlined here and I would welcome comments c/o the editorial address.

The main purpose of this article has been to discuss the use of computers in caving, avoiding applications such as membership lists and wordprocessing. However, research unpublished is worse than research undone, so readers may be interested in how easy it is to produce high quality reports, such as this journal, mainly using a home computer.

The typing, proofreading, correcting and editing have all been done with just a basic BBC micro, using Acorn's popular VIEW wordprocessing software, a single disc drive and a cheap printer. The final formatting and printing were carried out over two weekends using some rather more expensive kit. The cost of renting such equipment would not be beyond the reach of most clubs producing a journal such as this. It would only be needed for a short time, provided most of the work had been done in advance. The final copies to go to the printers were produced on a Apple Laserwriter using a pre-release copy of ViewPS, which runs on the same BBC micro using the files prepared with VIEW.

References :