Explains some of the science and technology behind GPS, and how to avoid errors.
This guide hopes to explain some of the reasons why readings can be so different, and how you can collect the most accurate locations possible.
Why can readings be so different?
The accuracy of a given reading will depend on many things – where you are standing in relation to the building, your orientation, the presence of trees and hills, the number and position of satellites available. GPS manufacturers claim that the accuracy of readings is in the order of 10 metres (radius), but this is ideal conditions. There are expectations of higher accuracy, but for mapping purposes 10 metres is all that is usually claimed.
The main problems with GPS locations are:
Wrong datum - A datum is a method of modelling the shape of the earth and of attaching the mathematical sphere to the centre of the earth. In Australia, we have used three different datums over the last 50 years. AGD66 was the first national datum, and stands for Australian Geodetic Datum 1966. Then in 1985 came AGD 84, and in 2000 we changed
to GDA94 – Geodetic Datum of Australia 1994. This last datum moved Australia up to 200 metres to the north-east, with respect to the rest of the world. The exact amount of shift depends on where in Australia you are. GDA94 is, to all intents and purposes, the same as WGS84 - the global datum used by the satellite network.
This means that a measurement taken in AGD84 will be up to 200 metres different from one taken in GDA94. You may leave the unit set to WGS84, or have it set to AGD to match your printed maps. Either way, it is essential that you keep a record of the datum you are working in with your GPS positions.
Multi-pathing – Signals bounce off building walls, mountains, trees and any other obstruction. This causes interference and will reduce accuracy, often significantly.
Spread of satellites – Ideally, there will be one satellite directly overhead, and at least three spread evenly around the horizon. If all the satellites you are getting a good signal from are in the same quarter of the sky, there is little
opportunity for the mathematics to reduce errors.
Number of satellites - Most units will tell you if there are insufficient satellites to be able to fix a position. If you have an older unit, you will need to go to the satellite page prior to recording, and ensure that there are at least 4 satellites with a strong signal strength available. Less than four, and there is no way for the device to calculate the errors and reduce them.
Age of GPS – newer devices have more sensitive computer chips, and are less sensitive to spread of satellites or
multi-pathing. They are also faster to startup, and to acquire a fix.
What do we do about it? -
The Four Golden Rules
The best guide I have found for understanding and using a GPS is published by the NSW Department of Lands. It can
be downloaded from here.
The document is over 100 pages, but certain sections are particularly important. These are accuracy and how to test
what is happening. The more places you play with the gadget, the more you understand it. Read this guide, and always
follow their 4 Golden Rules.
Rule Number 1. For a 3D position, use four or more satellites. Do not use if only three satellites are available.
Rule Number 2. Do not use the location if the spread of satellites is poor.
Rule Number 3. Use and record the correct map datum.
Rule Number 4. Use averaging when you can.
The following notes should be read as a supplement to the Lands document.
Recording Waypoints (such as a hut location).
The topographic maps of NSW are produced at 1:100K and 1: 250K. Reading a position from the map borders is necessarily not very accurate, and the current KHA list may err by up to 100m. Nor are all sites published.
Always use averaging over about 10 minutes. Newer handheld units will do it for you, but with older units you may have to record ten values (one per minute), and do the calculation manually.
The best place to be to obtain the most accurate possible reading is on the roof. Not only are you in the middle of the location, but this would eliminate interference from signals bouncing off walls and trees, and provide the best possible view of the satellites in the sky. But this is not always possible. So then what?
If the location is in a clearing, take positions at each corner of the building. The satellite reception will vary depending on the direction of the satellites and the direction you are facing. Averaging the four corner values will help eliminate problems, and place the position in the centre of the building.
It may not always be possible to get a GPS location for a site. Just do the best you can, and record everything – no. of satellites, spread (if reported), and a sketch showing how you did it. The more we have to work with, the better our final location will be.
No matter how or at what speed you are travelling, the follow notes apply.
As you move, the reception is constantly changing. This is due to both the movement of the satellites, and to changes
So there will always be sections of the track where accuracy is considerably reduced. There is no way of removing these errors, unless you have a professional survey team and many days. This is unavoidable, and has two implications for any map tracks made with GPS points:
1. the track will be close to the same for all GPS users;
2. the map can be considered a guide for travelling, but not an accurate representation of where exactly the track
Most GPS units use only the satellites for altitude readings. The vertical accuracy using this method is about 50 metres, which makes it useless. Some units provide barometric altitudes which works in conjunction with the GPS height to provide an height value. Air pressure decreases with increasing altitude, but also changes with the weather, so regular
calibration against a known height is essential. This can increase the accuracy to about 3 metres.