Tony Heher[1]

tony.heher@gmail.com

This is an extract from an article published in the 2008 Journal of the Mountain Club of South Africa (MCSA). The full article is available here and an animated slideshow showing how a GPS works and the source of errors is available here.

Updated 2014 to reflect some of the recent developments with smartphones and OSM maps

Update 2020 with further smartphone and WhatsApp developments

The availability and performance of GPSs[2] is leading to increasing use by walkers and climbers. Praised by some and heavily criticised by others, a series of tests was undertaken to objectively determine how useful they are for mountain navigation. The simple answer is that yes, they can be useful, but can also give incorrect readings, and users need to know when and why. Besides its use as a hand-held position device, GPS technology provides a variety of mapping tools - for both those who have and those who don’t have a GPS - which are changing the face of mapping. The rapid adoption of smartphones, all of which have GPS receivers, is also making GPS technology widely available. There are also an increasing number of countries putting up GPS satellite systems [3] so there are an ever increasing number of satellites available which is steadily improving accuracy

“Set an endpoint, enter GOTO and follow the directions” is the popular misconception of what a GPS can do. The well publicized stories of motorists driving into lakes or down railway lines highlight the hazards of trusting blindly to the vagaries of electronic navigation. So what about their use in the mountains? To answer requires an overview of how GPS’s work and what makes up the “GPS package” as it is more than just a positioning device. This is necessarily brief – the interested reader is referred to the journal article for more detail and links to a number of sites which provide more information.

An important conceptual starting point is to consider how users find their way in the mountains. The navigation method used most of the time by most users is by having a “mental map”. This is knowledge in ones head of the topography, the paths and the route. It is most commonly learned by doing a route with someone who knows it or possibly from consulting a map and route description. Unlike a road, one does not get to a destination by following a GPS with turn-by-turn instructions at each junction! The important point is that the GPS should be an aid to constructing or filling out this mental map, adding what it is good at – a precise position – to what a person is much better at – an awareness of place.

GPS’s have been available and in use for mountain navigation for over 15 years, but the recent availability of electronic topographic maps is increasing their use because it is now much easier to record and plot data. Manual plotting on paper maps is a painful exercise and prone to error. When using the combination of an electronic topographic map and a GPS, the actual numerical coordinates of a position or track are an internal detail and there is no need to write down or even know what the latitude and longitude are for any position or track. (An exception is when conveying coordinates verbally e.g. in a rescue situation – this is discussed further below.)

There are also a variety of maps available which are integrated with GPSs. The basis for most mountain navigation in South Africa is the 1:50 000 topographical maps published by the Directorate of Surveys and Mapping[3]. Besides the well known paper maps, an electronic version is available for free (apart from the media cost) from Surveys and Mapping and all the commercial topographic maps are derived from this source. Three versions are currently available in South Africa (that I know of - there may be more): Garmap Basecamp (the update to Mapsource)[4]; Madmappers AfricaTopo[5] series; a GIS compatible set published by Planet GIS[6] and the OSM maps. The biggest conceptual change from paper to electronic maps is that the scale is now notional and maps can easily be produced to any desired scale: 1:5 000, 1:10 000 or anything else that meets the needs of the task at hand. This is enormously useful.

However, there are many other “maps”. The most useful is Open Street Maps (OSM) because it has offline street and contour maps for virtually every country in the world. It has a wide range of high quality maps available for use both on PCs and smartphones - for free. OSMand for Android, for example, has full contour maps of the whole of South Africa and in the Cape it shows all the main and many of the minor paths. It's a very useful tool. It's a must have for anyone with an Android phone. Another track logging and mapping tool that users have reported is useful is http://www.gpsies.com/, but there are many others.

OSMand also has maps available for virtually every country in the world. Going to the UK or Norway or Brazil or anywhere on holiday? Download the whole country before you go and you will have not only road maps, but also maps that show most walking paths - and contours. OSMand is much better than Google Maps for hiking as Google Maps shows only major jeep tracks and the like, whereas OSMand has both major, minor and even obscure paths.

But an important point is that once one has a GPS track (which is simply a list of coordinates 10-20m apart) one can use it on any of the maps – topographical, Google Earth, or any other. More importantly, one does not even need a GPS to use and display the tracks on maps. The tracks could have been obtained from friends or downloaded from www.mountain-meanders.com or other similar source.

So a GPS package (unit and maps) can be used for a variety of functions, including:

• Planning routes, including tracing paths in Google Earth and transposing these to a map or vice versa.

• Drawing maps of whatever area and scale is needed, based either on downloaded or shared tracks or paths traced in Google Earth or from personal knowledge.

• Keeping a record of outings (where, time taken, distance, key features, etc). This is particularly useful when revisiting an area as it provides an accurate reminder of the previous visit.

• Sharing routes with friends or on sites like www.mountain-meanders.com

• Finding the way. This is probably the least used, but useful when needed.

• Reporting a rescue location. This is of value to a rescue party provided the coordinates are accurately reported; including the coordinate system in use[7]. With the widespread advent of smartphones, even reporting coordinates is unnecessary. If the rescuee simply share their current track to the rescue party, they would know not only where they were, but where they had been, which could be enormously useful. And in WhatsApp there is a very useful function of sending a WhatsApp location pin. Simply do Attach/Location in a WhatsApp message and your location will be known to within a few meters.

And for the first two functions you don’t even need a GPS. One fallacy can be laid completely to rest - GPSs do not replace maps! They complement and facilitate the use of maps to a far greater extent than paper maps alone and the growing use of GPSs and associated mapping technology will stimulate a greater awareness of maps and skills in using maps – skills that seem to have atrophied in many mountain users!

More important than the technical details of individual functions is that, taken together, they contribute to establishing or reinforcing that all important “mental map” of a route that is the single most important resource for a mountain user. Having done the planning and looked over a route, in most cases a GPS (if used at all) is switched on and forgotten about until one gets home and downloads the track one has walked for record purposes.

Given these benefits, what’s the problem? A number of users noted anomalies in readings that jumped hundreds of metres off a known track (or even kms in some cases). This was taken up by the GPS User Group who undertook a series of tests of a wide range of different GPS units, either owned personally by the users or on loan from the local Garmin agents, Avnic Trading. The tests consisted of walking well known routes with all GPSs simultaneously recording the same track and comparing results.

Some initial tests comparing different smartphone GPSs and dedicated GPS units like Garmin, indicated that smartphone performance was worse than that of a dedicated GPS. But this has changed rapidly with improved chips in newer smartphones which receive signals form more than one satellite newtwork . However, observation of a track, rather than a single point, often gives a good idea of the track accuracy. If there is a lot of "jitter" one knows to interpret a point position cautiously. Even better, if a rescuee is sitting still and leaves track recording running for 20-30 minutes, the resulting track usually shows a "jitter" where an average gives a more accurate location.

The full results are too detailed to report here but are available here. Over a period of a few months dozens of tracks were compared with up to 8 GPSs in use at any one time. The results were consistent in any given set of conditions so there is confidence that the conclusions are generic and apply to any GPS is use in similar mountain terrain anywhere in the world. Interestingly, there were no systematic differences noted between GPSs of widely different cost. R1000 units performed just as well, and sometimes better, than R8000 units. The only consistent difference is that the new Garmin H-Series high-sensitivity receivers perform better under tree cover, but they are no better on steep terrain.

What are the typical errors encountered? There are three sources of error: inherent GPS accuracy, map registration errors and GPS “jitter” – for want of a better name.

The inherent quoted accuracy of most handheld GPS units is of the order of 10-15m. In practice, recording multiple tracks on the same path simultaneously with a number of GPSs, the typical accuracy is about 25m, which is entirely adequate for most mountain navigation. This is under good conditions with a clear sky view.

However, tests of maps at various locations where one’s physical location can be determined by the topography (a coast line, knife edge, river junction, peak, etc) indicates that the maps can be misaligned by 50 to 100m, with the error varying in size and direction within a few km. Google Earth exhibits similar errors, but at times in the opposite direction, so the discrepancy between Google Earth and the topo map may be 100m or more.

Are these errors significant? Although they present some difficulties for map users[9], in practice they are minor for a hiker. Anyone silly enough to walk in the sea, or into space instead of across a knife edge, “because the GPS said so” would undoubtedly have killed themselves long ago! A GPS is an aid to navigation, not the gospel and of course must be interpreted sensibly in response to the actual topography.

The third source of error is more significant. This is what is called “jitter” and appears as rapid variations of 100m to 200m or more. An example of a test on the Grotto-Fountain-Cairn traverse illustrates this phenomenon. While walking along the pipe track and up the initial part of Grotto Ravine the multiple tracks, from the 8 GPSs that were used, correlate to within 25m. As one moved up the ravine, however, under the high cliffs of Fountain Buttress, the tracks diverged and “jittered” with a 100m-150m spread initially and later 200m-300m as Fountain Ravine was crossed. Once away from the high cliffs with a clear sky view the readings immediately settled down to the normal 25m spread.

This behaviour has been seen repeatedly where one is close to a high cliff and is especially bad where the cliff is south facing because more satellites are obscured in this situation. But it is not 100% repeatable! One person can walk along a route on a particular day and time and obtain a clear track, and the next person will not.

No description of or warnings about these errors is found in any of the GPS literature. The only mention is made of possible multi-path errors of a few metres due to reflection off rock faces. This is misleading. An analysis of GPS geometry and the orbital positions and speed (which is beyond this article – see here for an animated slide show) shows that the likely cause of the error is dilution of precision where many satellites are obscured (by the rock face) and those few satellites that can be seen are in close alignment so that the circles (or more precisely spheres) overlap tangentially rather than at or near right angles.

Because the 24 satellites in the GPS constellation tend to be located more towards the equator and the north, these errors are more likely when below a south facing cliff or in a ravine as fewer satellites are visible. But they can occur at anytime where the mountain topography impedes a clear view of the sky.

So, are GPSs useful in the mountains? Yes, they are but users need to be aware of the possible errors and be able to read a map … and the topography. They are an excellent aid to building or reinforcing the essential mental map and are also good for finding the start of a route or the top of a pass or a path on high open ground, etc. But they are not good on steep terrain, especially below south facing cliffs.

It is this latter limitation which contributed to the construction www.mountain-meanders.com the multi-media mountain wiki which combines GPS tracks with maps, photos and route descriptions for a set of resources which between them are good for documenting all types of routes so that they can be enjoyed safely and enjoyably by a wider audience.

[1] The assistance and contributions of many in the GPS User Group in both helping with the tests that were done and in commenting on this article are gratefully acknowledged.

[2] GPS – Global Positioning System. A network of 24 satellites and associated ground stations that enable portable handheld devices to determine their position anywhere in the world to an accuracy of a few metres. The term ‘GPS’ is commonly used to refer to both a handheld unit and the entire system. All modern GPSs also create tracks which are simply lists of successive latitude and longitude coordinates, typically 10-20m apart.

[3] Besides the USA system there are: GLONASS (Russian), Galileo (Europe) and BeiDou (Chinese). GPS receivers (which are embedded in smartphones) usually connect to the USA system, GLONASS and Galileo, so there are an ever increasing number of satellites available which is steadily improving accuracy

[4] https://www.garmin.com/en-ZA/

[5] www.madmappers.com

[6] www.planetgis.co.za

[7] Two aspects must be known: the grid system in use and the format of the latitude and longitude readings. While there are a large number of grid systems that can be used, WGS84 has become widely used and is the default in many GPS. But the user needs to know and check this is the case. More important, is to know what units are being used. Latitude and Longitude can be reported in Degrees, Minutes and Seconds; Degrees Minutes and Decimal Seconds or Degrees and Decimal Minutes. All three are used interchangeably in any GPS simply by selecting a display option. To report a position verbally, one must know which system is in use and how to read it. Fortunately this will disappear as in issue in the future as an increasing number of countries are requiring that the location of all emergency calls are automatically determined from GPS information in the phone. This is why most cell phones will soon have a GPS receiver in them – by law.

[9] The issue with the map and GPS track is what to represent, the reality or the error? Taking the Karbonkelberg path for example, if one plotted the real GPS track on the map it would imply that route was on the sea edge, but if one used a track that was correct on the map and then followed it with a GPS, the GPS would indicate a route that was 100m further inland than reality. “A GPS compatible map” is a fiction if, as is commonly the case, the map and GPS track are not in sync. The only practical solution to these sorts of (rare) circumstances, where the accuracy may be of importance, is to plot both tracks with a note that there is known to be an “ error of x at this point”.