FIG Peer Review Journal


OPUS-Database: Supplemental Data for Better Datum Conversion Models (4860)

Daniel Roman and Neil Weston (USA)
Dr. Daniel Roman
Research Geodesist
NOAA/National Geodetic Survey
1315 East-West Highway
Silver Spring
Corresponding author Dr. Daniel Roman (email: dan.roman[at], tel.: + 1 301-713-3202 x161)

[ abstract ] [ paper ] [ handouts ]

Published on the web 2011-03-16
Received 2010-11-22 / Accepted 2011-02-10
This paper is one of selection of papers published for the FIG Working Week 2011 in Marrakech, Morocco and has undergone the FIG Peer Review Process.

FIG Working Week 2011
ISBN 978-87-90907-92-1 ISSN 2307-4086


NOAA’s National Geodetic Survey (NGS) develops geoid height models (e.g., GEOID09, Roman et al. 2011) as tools for datum conversion from easily obtainable ellipsoidal heights to more practical orthometric heights. The official geometric model in the United States is the North American Datum of 1983 (NAD 83). This datum extends outside the Conterminous United States (CONUS) to other regions (e.g., the Marianas Islands); however, all are based on the same GRS-80 ellipsoidal model. Geopotential (orthometric) heights in all regions are defined in a consistent manner and tied to a datum point at a local tide station, such as Father Point/Rimouski for the North American Vertical Datum of 1988 (NAVD 88). For surveyors, it is more desirable to have a geoid height model based on these datums to provide a ready conversion tool away from the locations where bench mark control sites exist. To that end, GPS-derived ellipsoidal heights on leveled bench marks (GPSBM’s) are extracted from the NGS Integrated Database (NGSIDB), then least squares collocation is used to determine the separation between the datums between the points. While over 20,000 points were extracted for GEOID09, some of these were rejected due to suspect data and most were poorly distributed. Nearly 4000 of the total points were in one state and almost half were from a handful of states. This left the bulk of the country poorly covered and increased the likelihood of significant (cm- to dm-level) interpolation errors. Given the length procedures required to enter data into the NGSIDB (“Bluebooking”), surveyors have been reluctant to enter more information. Hence, only about ten percent of the nearly 500,000 benchmarks across CONUS have actually been occupied. To help mitigate this paucity of data, the Online Positioning User Service Database (OPUS-DB) was explored as a source for supplemental data. This nascent database is rapidly being accepted by the broader surveying community and can even be used to target significantly deficient areas. A pull of OPUS-DB in November 2010 yielded 422 points of which 80 were used in the making of GEOID09. While this is small in comparison to the overall NGSIDB data, the potential for growth is significant. More so is the fact that the data that the 342 point that were not used in GEOID09 show good distribution across the country and can potentially be used in future models to reduce the interpolation error caused by significant gaps in the control data coverage. The OPUS-DB data were generally noisier (standard deviation of 3.1 cm (one sigma)) than that from the NGSIDB (standard deviation of 1.5 cm (one sigma)) when compared on existing benchmarks with GEOID09 and NAVD 88. However, the OPUS-DB showed similar levels of noise at the supplemental bench marks not used in making GEOID09 (SD of 3.7-4.4 cm (one sigma)). This is consistent with the level of agreement seen between ellipsoidal heights between NGSIDB and OPUS-DB (SD of 4.4 cm (one sigma)). OPUS-DB demonstrated very good agreement with more rigorous Bluebook data and potential for use in future geoid modeling.
Keywords: GNSS/GPS; Positioning; Reference systems; heights; GPS/Leveling; geoid