Low-cost GPS surveying for amateur field studies groups

Low-cost GPS surveying for amateur field studies groups

The Golden Valley (Landranger Sheet 161 SO3141 to SO3927), lying in the eastern foothills of the Black Mountains west of Hereford, has been the scene of intense human activity since the late Stone Age. Several amateur history groups are engaged in exploring and recording the visible sites and alignments of landscape features for possible professional investigation at a later date. One significant area of interest is the indications of a Roman cavalry patrol route from Abergavenny to Hay-on-Wye, engineered with anti-ambush features and dating from AD 50-51, when the Romans were pushing westward along a line broadly similar to the modern A40 trunk road. Another group is investigating the remains of a very ambitious 18th-century land drainage and irrigation scheme, developed by a local landowner with utopian dreams for improving the prosperity of the area.

The survey requirement is therefore to record the positions of many fairly insignificant features in the hope of interpreting their locations into a broader picture and, hopefully, to provide a permanent record to guide future investigators. In some cases, where a site sketch survey is to be made, a secondary requirement is to lay a datum line across the site in a known orientation. In numerical terms the objective is to be able to determine spot location to an accuracy of 1m or better in terms of its British National Grid position. Such a requirement would be trivially easy to achieve using well-established GPS survey techniques, but the cost of the requisite survey-grade equipment is completely beyond the means of unsupported amateur groups.

Three developments during the past year (2000) have, however, completely transformed the prospects for low-cost GPS surveying:

• The removal, by order of the US president, Bill Clinton, in May 2000 of the deliberate degradation in positional accuracy, Selective Availability. The result of this development is not only to reduce the errors of raw GPS position information to the order of 5-10m, but also to make the error statistics much smoother and better conditioned in their effect.
• The change in policy by Ordnance Survey in making available via their Internet web site the raw GPS (Receiver INdependent EXchange format - RINEX) data recorded from the active stations of the UK National GPS Network. Public access was also granted to the catalogue of precise coordinates of the triangulation pillars incorporated in the network. The ready availability of these two classes of data brought the possibility of calculating the precise coordinates of user sites through post-processing.
• The discovery of undocumented features in the software of the inexpensive Garmin GPS12 family of consumer-grade receivers, by which raw GPS data could be output to a PC, converted to RINEX format and then post-processed against the GPS data from the active stations. The results of this detective work are available in the user-friendly and modestly-priced GRINGO package sold by the IESSG at Nottingham University. A companion program performs the post-processing functions. Full details of GRINGO and a limited function demonstration version can be obtained from the IESSG web site at: http://www.nottingham.ac.uk/iessg/gringo.

Thus, the user on even a quite limited budget now has the means to achieve much better levels of survey accuracy than have hitherto been achievable.

Over the past winter a quite extensive range of familiarisation and evaluation trials have been performed.

Firstly, to examine the observing and post-processing tasks a Garmin GPS12XL receiver attached to a laptop PC running the GRINGO software was positioned at the triangulation pillar on the summit ridge of Blakemere Hill. The processed position (using RINEX data from the Ordnance Survey Droitwich reference site, some 50km distant) was compared with the precise coordinates published on the Ordnance Survey web site. It was quickly established that positions could be measured to within about 50cm mean error, albeit the standard deviation of the data points collected during a one-hour observation period was of the order of 2.5m.

The second phase of the trials aimed to repeat the exercise at a reference station to be established at the home of the author of this report, where a GPS receiver could be connected to a desktop PC. Although the observing conditions, with the receiver propped in a south-facing window, were far from ideal, mean positions within a 70cm diameter circle and consistent with the coordinates read from a 1:500 scale local planning map could be obtained, provided that some care was taken to exclude data points taken at times where one or more satellites were in the process of rising or setting behind the nearby ridge of hill that obscures the sky view up to some 20 degrees elevation. While such an event is in progress the processed differential GPS position makes very large excursions (c 500m or so) along a line parallel to the direction of the horizon-obscuring ridge.

On the basis of these trials, the purchase of an additional GPS receiver can now be justified. The object will then be to use a differential GPS technique, operating over a distance of only a few kilometres from the local reference station.