Previously we discussed the benefits of using DOP as a planning tool. However, DOP has its limitations.
While a poor DOP will generally mean poor accuracy, a good DOP may not mean good accuracy.
With older receivers, as well as high accuracy receivers, if a satellite is bring tracked then generally, we can make a decent measurement. With new high sensitivity chipsets that now come in many consumer and entry level GIS handhelds, satellites are tracked even when deep within a building. This means that the DOP value can be good, but because of systematic errors such as multipath, the position error is very poor.
Why is this a problem?
Most experienced GPS users have the skills to know that buildings, trees, cars etc. cause multipath that can lead to very poor accuracy. But many introductions to GPS discuss how a low DOP value is a good way to determine if your position is reliable. For example Wikipedia has this example:
The plots below shows the difference between a "high accuracy" or survey grade chipset, and a commercial "high sensitivity" chipset. Both devices were placed on my desk,within my office.
While the survey grade chipset is unable to reliably track any satellites, the high sensitivity chipset is tracking 8+ satellites and reporting a DOP of less than 2.0 even though it is within a concrete building with few windows!
If we look at the position plot of the high sensitivity receiver the position is all over the place even though the receiver was motionless.
Although the above case is an extreme example, there are lots of occasions where a user will encounter high multipath environments, especially for GIS applications. With the latest developments in high sensitivity chipsets inexperienced users who rely too much on DOP as an indicator for the accuracy of the solution may be in for a surprise when they look at the results afterwards.
How can we solve this problem?
While it is possible to improve the reliability of DOP by trying to account for systematic errors and properly modelling the covariance of the observations most users cannot implement this approach. This is why experience and training is still essential for proper data collection even for GIS applications. Although GPS receivers have become incredibly common, and manufacturers continue to improve user interfaces to make it possible for even novice users to jump right in, there is still a need for proper training, education and user experience to ensure good quality data collection for your projects.