Saturday, January 7, 2012

A False Sense of Confidence

For Christmas my lovely wife gave me a new GPS receiver.  At work I'm surrounded by some of the most expensive and precise GPS surveying technology available today - devices that can reliably locate you to within a few centimeters of your true position.  What I didn't have was a consumer grade GPS device that I could take out hiking or fishing.  It seems Roberta read my Dear Santa letter and got me a DeLorme PN-60 receiver.

Many of you are probably asking, "why not just use your iPhone?"  Good question.  While I love my iPhone experience has taught me that it makes a lousy GPS receiver.  It may locate you good enough to allow accurate vehicle navigation while cruising up the interstate and it may provide good enough locations to allow you to update your Facebook status, but as a dedicated GPS tool it just doesn't cut it.  To be fair, the iPhone does excel is as an integrated device for wayfinding.  Apple's integration of hardware and mapping software makes the iPhone a wonderful and reliable urban navigation tool.  However, most of this functionality is dependent on a wi-fi or 3G signal, making the device unreliable for use in the backwoods.

What I wanted was a dedicated GPS that is rugged, water resistant, accurate and allows me to upload maps and aerial photos.  The PN-60 seems to fit that bill just fine.

Of course the first thing I had to do is take it out and play with it!  A few days after Christmas I loaded the dogs into the old Volvo wagon and we headed out to one of my favorite parks with walking trails.  This park, Lake Horton, is located in Fayette County and has a number of walking trails that skirt the lake.  Once we got there I leashed up the dogs and set the PN-60 to generate a track file (basically recording my location every 10 feet).  The trail we walked is a 1.8 mile long trace that moves through a mix of topography - some open ground, some heavy canopy cover, some areas where the canopy obscures just half the sky.  It is a good test environment for a GPS receiver.

Let's pause for a minute to review just how GPS works.  The GPS receiver establishes a position by a triangulation process that uses signals from the GPS satellites in orbit around the Earth.  A GPS receiver needs good signals from at least three GPS satellites to calculate a useable position in 2 dimensions (horizontal).  Note that the key word here is useable, not accurate.  To get a useable 3 dimensional position (throwing elevation into the mix) the receiver needs good signals from at least four satellites.  The accuracy of the position your receiver calculates is highly dependent on a number of factors such as satellite geometry (where they are in the sky overhead), how healthy the satellites are (yes, there are 'healthy' and 'unhealthy' GPS satellites), the amount of overhead interference (things like heavy tree cover, tall buildings, etc.) and the overall sensitivity and accuracy of the receiver itself.  All of these factors have a vote in how accurate or inaccurate you GPS location will be.  The only fixed variable in this equation is the capability of your GPS receiver; all other factors can vary from minute-to-minute and location to location.

Just how accurate can a GPS receiver be?  The US Air Force, which operates the GPS system, states that civilian users can expect the GPS system to provide position fixes to within 100 meters 95% of the time.  That's pretty darned conservative and assumes a very basic early generation receiver that is not applying any modern GPS correction signals from systems like the WAAS satellites (Wide Area Augmentation System).  

The civilian GPS industry tosses around the number 15, as in 15 meters.  A modern consumer grade GPS receiver operating under open sky and in the static mode (not moving) should provide a fix to within 15 meters (49 feet).  If the receiver is WAAS-capable (and most are) then that potential accuracy should drop to about 3 meters (10 feet). This is about the best a modern consumer grade GPS unit can be expected to do.

This big improvement in accuracy has come through improved receiver hardware and software.  GPS receiver manufacturers have gotten pretty good at this business over the past 20 years and they've developed some nifty tricks to greatly improve overall accuracy and satellite signal acquisition speed and lock.  A modern GPS receiver, even a relatively inexpensive recreational unit like the PN-60, is a marvel of modern hardware and software design.

Now back to our regularly scheduled blog post...

So the dogs and I walked the 1.8 mile trail and then rushed home to upload the GPS track file to the DeLorme Topo North America desktop software to see what was recorded.

Zoomed out it looked sort of OK; at least I was on the right part of the world.

DeLorme Topo North America 9.0 (click on image to enlarge)

But when I zoomed in to specific sections of the track I noticed some concerning issues with how the recorded GPS track was lining up with the portions of the trail visible in the image.

25 ft. offset between the walking trail and the GPS track
(click on image to enlarge)

Note the high tension lines running north-south across the image.  This is a potential source of interference.


38 ft. offset between the walking trail and the GPS track
(click on image to enlarge)

Note the heavy canopy coverage immediately south of the trail.  Heavy tree canopy can either block or interfere with GPS signal quality.


An average 10.2 ft. offset on a section of trail with light overhead canopy
(click on image to enlarge)

In an area with fairly open sky (and well away from powerlines) the average offset drops to just over 10 feet, perfectly acceptable accuracy for this receiver operating in the moving mode.


Offsets ranging from 14.5 ft. to almost 25 ft. in an area of completely open sky
(click on image to enlarge)

What's going on here?  This is an area of completely open sky - no trees or other obstructions.  Accuracy should be much better than what we are seeing here.  Well, if you look closely you can just make out the trace of the high tension power lines cutting across the lower right hand corner of the image.  My suspicion is that the powerlines are interfering with the GPS signals.

And last:

Two sample areas in close proximity - one showing a 60.7 ft. offset and
one showing almost negligible offset
(click on image to enlarge)

This example clearly illustrates the impact of heavy tree canopy cover.  The upper measurement, taken in an area of dense canopy cover, shows an almost 61 foot offset between the walking path and the GPS track.  However, just 55 feet to the south, under fairly open sky (and with no power line interference) the offset is so small as to be negligible.

So what can we learn from this simple example?  Let's review:

1. The manufacturer's stated GPS accuracies are for their products operating under the most ideal conditions - open sky, with good satellite geometry and no electromagnetic interference.  Your real world results can, and will, vary widely.

2. Overhead or adjacent obstructions like trees can have a dramatic effect on GPS accuracy.

3. Unseen influences like nearby powerlines can have significant impact on GPS accuracy.

4. In areas of little obstruction or interference consumer grade GPS receivers can achieve their claimed accuracy of +/- 3 meters (10 feet) in the moving (tracking) mode.

So what can you do to improve your GPS receiver's accuracy?  Here's a few tricks:

1. Make sure WAAS is turned on.  On most receivers sold in the US it is turned on by default, but you still need to check.

2. Make sure position averaging is turned on.  While this won't help when moving, when stopped to record a position (waypoint) if you let the receiver average your position for a few seconds before storing it your accuracy can improve dramatically.  When collecting waypoints I'll try to let the receiver average between 5 and 10 seconds before storing the point.

3. Hold or carry the receiver so that the antenna is fully and properly exposed to the sky.  Don't expect a GPS receiver buried deep inside a backpack or tucked inside a shirt pocket to return a good position.  Your receiver has to 'see' the GPS signals and anything you can do to improve it's view of the sky will result in more accurate positions.

Understand your potential sources of error!  Remember how the system works - radio signals from space.  If you understand what can interfere with those signals and you remain observant of your environment you can often work around these sources of error.

So many people I talk to about GPS have a false sense of confidence about GPS accuracy.  The vast majority think their receiver is far more accurate than it really is.  While the GPS satellite system and our modern receivers are true marvels of modern technology the system still has plenty of real world limitations.


Note: I do have to add that the measurements I show in the images above also have a lot of built-in but uncontrollable error.  The DigitalGlobe images I'm measuring against have an advertised resolution of 30 centimeters (roughly 1 foot).  Assuming these images were processed without a tight orthorectification their inherent positional error can range from 2 - 5 feet (my estimation).  In this blog post I'm not trying to establish absolute accuracies or inaccuracies for a given receiver.  I just want to give the reader as sense of what impact physical and electromagnetic interference can have on GPS accuracy.


Anonymous said...

I don't know, Brian. I think your new GPS tracked your path just fine. My conclusion about why there is deviation from your track file and the path shown on the map is that you were weaving about in the park after too much post-Christmas egg nog!!! Perhaps you need a guide dog to keep you true!

Brian said...

Hah! The more likely explanation is that I had three dogs dragging me around. Actually, I didn't mention it in the blog but I took great pains to stay on the center of the trail all the way around. This was actually the second track I ran at Horton. The first was a few days earlier and I got pretty much the same results. I did this second track as a check and to make sure I didn't have any incorrect settings on the GPS receiver.

Bill said...

I'd try my 5+ year old Magellan as a reality check but I don't have the downloadable goodies for a real comparison. However, mine will realiably locate a secret fishing spot on an unnamed creek and if I can get within a few meters I'd be happy.

Brian said...

Bill, you are right - it's all a matter of perspective. While we stand in a spot with GPS receiver in hand and get all steamed about it not accurately locating us any closer than 10 feet we forget that not too many years ago, working with 1:50,000 scale maps and magnetic compasses all we had to do was get to within 100 meters to be considered good to go.