Sign in to follow this  
LilBudyWizer

Datums, Mapping and GPS, Oh My!!!

Recommended Posts

LilBudyWizer    491
I have a GPS logger. The software I have is rather limited in the processing and summarization of the data. It also isn't well documented as to what it does with scrubbing the data. What I'm trying is using WGS84 to convert the longitude, latitude and elevation to rectangular coordinates. To do that I find a vector to the WGS84 surface, call it v, then scale it by (1+elevation/|v|). If I don't scale for the elevation then I get within about 2% of the total distance the software produces. Due to the scales involved and the fact my test log has 2400 samples I'm inclined to think that error can be accounted for by numeric errors. One problem is the elevations. They seem to be within only about 100 feet and very irratic. The position is only suppose to be within 10m, but it drifts slowly so it seems far more accurate from reading to reading. I can certainly recognize the path and detours I took while walking. A profile of the elevations is a whole differant story. I wouldn't recognize it as what I walked at all. So one thought is to use the USGS Quads to push it to the ground. I'm wondering if anyone has any experience with dealing with GPS data and could offer tips, suggestions or caveats. Particularly concerning elevation. I have ideas on how I want to deal with the lack of accuracy for the horizontal data, but I'm not real sure how to deal with the elevation data. I hate to completely discard it and replace it with the USGS Quad elevations. I'm more interested in relative than absolute elevations. The irratic nature of it makes it seem about useless though.

Share this post


Link to post
Share on other sites
etothex    728
I have worked with GPS before, and even on the high accuracy units the elevation/altitude reading sucks majorly compared to the horizontal. Don't rely on it for any kind of absolute or relative accuracy, if there is a feasable alternative.

One of the bigger things that affects altitude accuracy is crummy satellite geometry. If you have all your sats within 30 degrees of horizon, then you get crummy altitude readings. You can check this by looking at the GPS sat geometry data as well - unfortunately there's not much you can do to fix that error besides finding clear sky and hoping for good geometry.

how are you getting that 2% figure? relative error? (i.e., move precisely 100m and the gps reports you moved between 98m and 102m?)

I dunno much about this part since someone else wrote the software to do this but I know he wrote it so the only inputs were lat/lon, and we got very, very accurate UTM rectangular coords out of that. But if you want I can look back and see what I can dig up...

Share this post


Link to post
Share on other sites
Prospero999    122
Hi,

When you say you're trying to convert to rectangular coordinates, do you mean ECEF (Earth-Centred Earth-Fixed)? If so, I can dig out a good guide in pdf format which has all the relevant formulae. I'm new here so I'm not sure how/if I can attach the file...

Anthony

Share this post


Link to post
Share on other sites
LilBudyWizer    491
Street Atlas 2006 gives a summary of the log, i.e. sum, min, max, avg, start, end and elapsed. So I get within 2% of the total distance it reports when processing the log. Generally I would think I should be able to match to a higher degree, but I lose a lot of digits of precision due to scale. The radius of the Earth is about 20 million feet and I'm moving 10 feet on average per reading so it seems I would be losing six digits right there. It does seem a bit high of an error for rounding so I wouldn't rule out a differant datum.

I'm not converting to a mapping coordinate. Rather just from spherical to rectangular for the WGS84 oblate spheroid. The software will plot the tracks for me so I don't have any real need to plot it on a map myself. If I try using the USGS Quads to foce the tracks to the ground, which is starting to seem the best choice, then I'll have to convert to their coordinates. As far as uploading files this site only provides space for GDNet+ members as far as I know. If it's text you can paste it into the post. The FAQ explains how to use the source tags so it puts it into a scroll box. You can also go into edit on any post, though you can only actually update your own, to see the tags they used.

I know the Topo USA will display the dilution of precision information when connected to the GPS. I'm not sure the handheld software displays it though. If it does though it might be a good idea for me to start paying attention to it. So far the status information has been pretty much gibberish to me, but perhaps recording some of the status with waypoints might make it make a bit more sense to me. Overall I hadn't really thought about recording known points. That might be a better starting point that dealing with the entire logs.

Share this post


Link to post
Share on other sites
LilBudyWizer    491
I guess I really need to just play with this data a bit more. Rather than starting with tracks I perhaps should start with fixing points. If the receiver is stationary for a period of time the logged position drifts. So how do you fix a point accurately? Particularly is there basically a "good enough" sample interval and number of intervals. Also are the GPS satallites geo-sationary or do they move across the sky? If they move is there a way to tell when optimal times are for using GPS in a given area?

Share this post


Link to post
Share on other sites
LilBudyWizer    491
The differance was a rounding error. I found the format for the binary files and it's far more accurate. I match the software to the precision it reports in now using mean sea level. I also found how to detect when I'm stopped. Apparently the GPS estimates instantaneous velocity. Just as an example the distance between log points two seconds apart was 9.74 feet which is about 3.32mph, but the recorded speed was about 0.3mph. I still have to play with it, but I think I may be able to use the reported speed to clean it up in general.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

Sign in to follow this