Talk:Import/NGS OPUS Control Points

Accuracy / Coordinate System

I don't know if this project is still ongoing... I have actually been adding NGS control points to OSM, on and off, for a long time. I just want to note that the positions provided by the NGS are in NAD83(2011) at the reference epoch of 2010.0.

The current "realization" of WGS84 is G2139 (equivalent to ITRF2020 at the reference epoch of 2015.0).

The point of this is that a simple "coordinate system transformation" from NAD83(2011) to WGS84 will give positions that are up to 2 meters "off" from ITRF2020 at the reference epoch, as well as from WGS84 at the "imagery" epoch. This means that in order to preserve the accuracy inherent in the data you are importing, you should probably explicitly state the datum and epoch of the data. Hopefully, someday OSM will itself account for data epochs.

When I have been adding NGS benchmarks, I have been using NOAA VDatum (https://vdatum.noaa.gov/vdatumweb/) to explicitly transform the coordinates from either NAD83(1986) (the reference epoch is 1984.0) or NAD83(2011) (epoch is 2010.0 or as stated on the datasheet) to ITRF2020 at the reference epoch of 2015.0, and then explicitly stating "survey_point:datum=WGS84 (G2139)" and "survey_point:datum:epoch=2015.0". VDatum explicitly includes corrections, btw, for continental drift.

Using VDatum also allows for a highly accurate correction from the NAVD88 vertical datum used by the NGS, to the EGM1996 datum used by OSM. It's probably worth pointing out that EGM1996 is not the same as an ITRF2020/WGS84(G2137) elevation... for (as an example) WILDER RESET 1930 1890 = CN0068, the NAVD88 elevation is 219.375, EGM1996 is 218.997, and ITRF2020 is 189.495. To get EGM1996 elevations out of VDatum, you have to explicitly select it as the target vertical reference frame, which will then give coordinates in the (obsolete) WGS84(G1674) horizontal frame. The software will ask for the source and target epoch when making this transformation, but they are irrelevant as long as you only use the elevation (the difference between NAVD 88 and EGM 1996 is a constant at any given position).

This may all seem very nitpicky, but in cases where the benchmark (or it's monument) is actually visible in satellite imagery, the positions in an NGS shapefile opened directly in JOSM (i.e NAD83(2011) to "generic" WGS84) are visibly "off" from the imagery, while the positions given by VDatum after the epoch shift land almost directly on the object. Revent (talk) 18:37, 6 November 2023 (UTC)