[aprssig] Accuracy of data in GIS databases

[Jim Lux]

Wes Johnston wrote:

There's a law here in SC that requires that all data in the GIS system be
collected by certified GPS surveyors (something to that effect).  The problem
arose when a city tried to get the surveyors to locate all the fire hydrants.
They could not get them in for some length of time... and decided to recruit a
boy scout troop to run all over town with hand held GPS units and mark the
hydrants.  The surveyors responded by sueing the town to have the data removed
from the GIS system.  What a crock of poo poo.  One of the higher ups at the
state GIS level told me this story about 2 weeks ago when I was discussing
getting the fire hydrant locations for my town's fire trucks.  We were going to
raise a little money and donate cheap hand held GPS units to the volunteer fire
dept's... they can "goto" nearest waypoint when on a call if all the fire plugs
are saved as waypoints.  To my way of thinking, they could have derated that
data in the GIS system, but still kept it until it would have been superceeded
by the certified surveyors' data.  Some data is better than no data.


---

Allow me to present a somewhat different viewpoint.

Some data, of uncertain validity, is worse than no data.
There's a lot you didn't say in your story. (and I was unable to find out 
anything more by googling for it)

I'll be the first to admit that some of the "practice act" restrictions on 
engineering and surveying smack of protectionism (full disclosure: I am a 
registed P.E. in California), however, when you're in the public safety and 
regulatory business (which the city is), a bit of professional advice might 
not be amiss.  There are a lot of subtleties in the whole "establishing a 
position" thing, particularly in connection with GIS database, that aren't 
obvious.

Consider standard USGS maps.  They have a statement "meets national map 
accuracy standards" on them.  What does that mean in the real world?  It 
means that things on the map are within 1/50th of an inch (a pencil point) 
of where they should be, which seems a reasonable requirement.  The 
resolution of the data is comparable to the resolution of the medium in 
which it's reproduced.  When you get to GIS databases, though, it gets a 
bit trickier.  You can print out a map at 1:100 scale when the underlying 
data is only good to, say, 1:100,000. It's not enough for them to put 
"disclaimers" on everything, and for all we know, the city's GIS doesn't 
have the ability to associate a "measurement uncertainty" with every point 
(or, at least, for accuracies as low as casual consumer GPS 
measurements).  Most points on city surveyed maps are probably good to 
better than a tenth of a foot (at least, you hope they are, when they 
staked out your lot lines, right?), and there's a general expectation that 
on plat maps, plot plans, etc. that similar accuracy is maintained.


When public agencies produce mapping products, there are certain accuracy 
standards that they have to meet to avoid sticky liability issues. Once 
data is in a database, you can't control the uses of it.  For all you know, 
those hydrant locations might wind up being used to evaluate building 
permit applications as to whether you need to have a reserve water tank or 
fire sprinkler system. (Around where I live, distance from the hydrant 
determines what sort of fire protection systems you need to implement.)


There's a bit more to surveying (particularly in a legal sense) than just 
whipping on out with a survey grade GPS receiver.  It mostly has to do with 
data validation and a realistic assessment of the position accuracy you need.

Were all those boy scouts going to have some sort of validation process to 
insure that there weren't egregious errors in the data (maybe because the 
handheld receiver failed, or didn't get a good fix, or just simple operator 
error)? Were the boy scouts going to write down/store the dilution of 
precision information (or EPE) when they made the measurements, or were you 
just going to assume the "lowest common denominator" 30 meter scale 
accuracy for C/A GPS. 30 meters is about the distance from the hydrant in 
front of my house to the street behind me... not too useful for a firetruck 
if they go to the wrong street.

Who was going to reconcile the GPS measurements (in whatever coordinate 
system and datum) to that used in the GIS?  This can either be trivial or 
complex, depending on how the GPS and/or the GIS is set up (most GPS don't 
provide information in state plane coordinate systems, for instance).  It's 
not just a matter of recording in UTM and importing (since the GIS was 
probably built from documents and surveys dating back literally hundreds of 
years).

Was someone going to do a reasonableness check on all those collected 
positions (as in driving by and verifying that Hydrant #ABCDEF is actually 
at the position recorded?)

I won't say that you couldn't have a troop of boy scouts with appropriate 
inexpensive receivers go out and do the data collection, IF (and ONLY IF) 
you had good survey design, which would typically require someone who's 
living is doing good surveys (i.e. that registered surveyor required by the 
law).

It's a shame that they had to resort to a lawsuit, but sometimes, that's 
what it takes. Or, that's the only mechanism by which such things can be 
handled.

Jim Lux, W6RMK