On Thu, Aug 12, 2010 at 11:05 AM, Bryan pbryan.heidorn@gmail.com wrote:

If there is likely to be multiple coordinate systems can we have a column that defaults to wgs84 but may be reset to other systems?

I don't think it is likely, given phones and GPSs as sources. Printed maps could be another story, but who uses those any more? ;-)

Does someone have a table that could be given to data collectors for estimating uncertainty? The variables in this case are limited.

The the contributors to uncertainty from the phones and GPSs can be found in various references, including the "MaNIS/ORNIS/HerpNET Georeferencing Guidelines" (http://manisnet.org/manis/GeorefGuide.html), GBIF Best Practices for Georeferencing (http://www2.gbif.org/BioGeomancerGuide.pdf), and "The point-radius method of georeferencing locality descriptions and determining associated uncertainty." (Wieczorek et al., 2004; IJGIS 18-8 pp745-767). 1) device accuracy - this depends on the device and the conditions at the time. Best to either measure these empirically or use conservative estimates (1km for non-GPS phones; 30m for GPS and GPS-based phones in clear, unobstructed conditions, worse under cover - perhaps 100m in forests, for example) 2) device precision - this depends on the latitude, the coordinate system the device is using, and the number of digits of precision it reports. For decimal degrees to the nearest 0.00001 degrees in Woods Hole, the contribution would be under 2 meters. For nearest second, the contribution would be 39 meters. 3) datum - if the datum is unknown for some reason (generally an avoidable situation), the uncertainty contribution can be determined either with a conservative, worst-case scenario (1.6 km maximum difference between WGS84 and any other published datum as of 2004), or with a calculated difference between the most likely alternate datum at that location (NAD27, for the uncertainty contribution at Woods Hole is 42 m (Georeferencing Calculator; http://manisnet.org/manis/gci2.html, using Coordinate source="GPS", Coordinate System="decimal degrees", Latitude="41.5219444", Longitude="-70.6681833", Datum="datum not recorded", Coordinate Precision="exact", GPS Accuracy="0", Distance units="m").

So, the best case scenario is a GPS with a known datum using degrees decimal minutes to three digits of precision under excellent conditions, in which case the real uncertainty is probably on the order of 5 meters. The most likely scenario is a GPS with a known datum under normal conditions using decimal degrees to five decimal places, in which case the real uncertainty is probably about 15 meters. The worst case scenario is dismal.

For non-GPS phones you'd have to determine their accuracy in the field by independent measures. I wouldn't recommend them for precision work.

sent from my Droid

On Aug 12, 2010 2:04 AM, "Javier de la Torre" jatorre@gmail.com wrote:

+1 for bioblitz.

I will add that because Fusion tables use WGS84 for geometries it is important that we don't mix them. We will be relying on the spatial capabilities on fusion tables for data visualization.

Javier www.vizzuality.com

On 12/08/2010, at 03:52, joel sachs jsachs@csee.umbc.edu wrote:

...

On Mon, 9 Aug 2010, Javier de ...

---

tdwg-tag mailing list tdwg-tag@lists.tdwg.org http://lists.tdwg.org/mailman/listinfo/tdwg-tag