History of the North American Datums
Before a national triangulation network connected the various portions of the U.S., regional datums had their own origins and standards for latitude and longitude, usually determined by astronomic observations. The U.S. Lake Survey based their surveys on the observatory in Detriot. The U.S. Geological Survey based their surveys west of the 100th meridian on the station in Ogden, Utah, which had been surveyed by Lt. Walker in 1877. The coastal surveys of California were based on local astronomic observations, but when telegraphic longitude was determined at San Francisco's Washington Square station, it became the standard of longitude for the West Coast surveys.
In New England, the Harvard Observatory in Cambridge, Massachusetts was designated by the CGS as the "cardinal point of longitude" in 1845 and its position was determined by exchange of chronometers with Greenwich and by astronomic observation. In 1866 and 1867 the longitude of Calais, Maine and Harvard Observatory were determined telegraphically by exchange of signals with Europe over the first transatlantic telegraph cable (note 1). In 1869, the difference in longitude between Harvard Observatory and San Francisco was determined telegraphically (note 2).
An Act of Congress in 1850 established the American Meridian as the basis of astronomy and surveying. The meridian was defined as passing through the dome of the Naval Observatory in Washington D.C. The longitude of that dome was not well established, though, and was first determined through triangulation of the CGS, and later by determination of the telegraphic difference in longitude from the Harvard Observatory. The difference in longitude was firmly established in 1867 as 23m 41s.11 (5° 55' 16.65").
New England Datum (1879)
Over the years from 1840 to 1878, a triangulation network, the Eastern Oblique Arc, was extended from Maine to Georgia and was connected to the coastal triangulations. Adjustments were made as additional observations were obtained. In 1863, latitude had been computed from 20 astronomic stations between Maine and Connecticut and was standardized at Blue Hill, Massachusetts. The azimuth from Blue Hill to Copecut was the standard for orientation. In 1867, longitude had been determined telegraphically at Harvard Observatory as previously mentioned, and was approved as a standard in 1869. The Bessel 1841 spheriod was used for all CGS charts.
By 1879, the Eastern Arc had been extended to Georgia and the additional observations merited another adjustment. The station Principio at the upper end of the Chesapeake Bay was approximately midway along that triangulation, and in 1879 the CGS performed a least squares adjustment of the 58 stations of the Arc and established a standard position for station Principio. All other stations were referenced to that station. This became the initial "New England Datum" used by the CGS for their coastal charts. The datum was specified in terms of the position of Principio, the azimuth from it to station Turkey Point, and the use of the Clarke 1866 spheriod as the standard surface.
Subsequent adjustments were done in 1880 and 1884 as more stations were connected to the Eastern Oblique Arc and telegraphic longitudes were expanded. During this time, station Principio remained the standard fixed point of the datum.
U.S. Standard Datum (1901)
A second great arc of triangulation, the Transcontinental Arc along the 39th parallel, was completed in 1899 and finally enabled the regional datums to be connected. In 1901, the CGS established the U.S. Standard Datum. The CGS reduced the impact of the new datum on the many charts and positions that had already been published by basing the new datum on the existing New England Datum, keeping station Principio fixed. At the same time, a new triangulation arc along the 98th meridian was being extended north and south of the Transcontinental Arc. In order to have a standard reference point near the intersection of those arcs, the station Meades Ranch was selected as the new standard point, after its position relative to Principio and the New England datum had been computed.
The 1901 U.S. Standard Datum was then specified in terms of the position of Meades Ranch, the azimuth from Meades Ranch to station Waldo, and the Clarke 1866 spheriod. Five thousand survey points were incorporated into the datum.
North American Datum (1913)
By 1913, the 98th meridian arc had been extended to both Mexico and Canada, and Canada and the U.S. had jointly established the eastern boundary of Alaska at the 141st meridian, and work was in progress on the 49th parallel survey of the Canada-U.S. boundary. Canada and Mexico, having been involved with these geodetic collaborations with the U.S., were convinced of the merits of a continental datum and they adopted the U.S. Standard Datum, which was renamed the North American Datum.
Once the U.S. Standard Datum of 1901 had been established and perpetuated as the North American Datum, all subsequent triangulation work was fit into that network, and all closure errors were taken up by adjustments to the new work, leaving the stations of the 1901 datum unchanged. By 1925, it was apparent that the new work was requiring adjustment far in excess of what could be attributed to measurement error. A decision was made to perform a new adjustment that included all the completed first order triangulation. The work was divided into two parts -- the Western network, consisting of triangulation west of the 98th meridian, would be done first because minimal additional field work would be required before the adjustment could be done. The western field work was accomplished by 1927 and the necessary field work for the Eastern network was finished by 1931. The adjustments were largely complete by 1933.
The adjustments were the first to make use of Laplace stations to control azimuths and correct for deflection of the vertical. The datum of 1927 retained the position of Meades Ranch as the reference point, but its azimuth to station Waldo was changed based on the corrections from Laplace stations. The datum, instead of being oriented solely by the azimuth at one station, was now oriented by 175 Laplace azimuths throughout the network. The Clarke 1866 spheriod was retained as the best fit to the data.
By 1979, surveyors were again finding inconsistencies between the datum and their new work. New surveys, performed with electronic distance measuring equipment and other high-precision technology, were capable of 1:1,000,000 accuracy. But when new surveys were adjusted to fit into the 1927 datum, they were requiring distortions far in excess of what could be attributed to error. In some cases, the NAD27 network accuracy was found to be no better than 1:15,000, resulting in errors of tens of feet.
Anticipating that the future of surveying lay in satellite technologies such as satellite doppler and GPS, the NGS decided to adopt an earth-centered reference frame, in contrast to the NAD27 frame centered on Meades Ranch in Kansas. The standard spheriod was chosen to be GRS80, which had been computed as a best fit to the worldwide geoid. To determine the location of the earth's center, data from the Navy's doppler satellite ranging project was used and the datum was oriented according to the International Bureau of Time standard BTS-84. As initially defined, the NAD83 and WGS84 datums were essentially identical.
The data for all control points was computerized and a full adjustment was performed on all the historical survey data. Tectonic plate motion, subsidence, glacial rebound, and other factors were determined and accounted for. The entire process took 7 years and cost over $30 million.
Shortly after NAD83's adoption in 1989, additional investigation revealed that its origin was displaced by about 2 meters from the actual center of mass of the Earth, which could now be determined to within a few centimeters. Furthermore, the NAD83 datum was defined so that the datum remained fixed with respect to the motion of the North American tectonic plate, whereas other global reference frames, including WGS84, were defined to remain fixed with respect to global net plate motions and had been revised to agree with the most recent determinations of the earth's center and orientation. These factors, along with others, resulted in a difference between NAD83 and WGS84, which is small but significant for precision surveys, amounting to differences of a few meters in some places.
In addition, the scale of the datum was found to be different by 0.0871 parts per million from the true meter, with the effect that ellipsoidal heights were too large by about 0.6 meter.
Soon after the NAD83 adoption, GPS technology became widely available and provided ever more precise measurements. Discrepancies were again noted between station positions that had been determined through classical survey methods and the newer more precise GPS observations. In order to reduce those errors, the NGS, along with individual states, established the High Accuracy Reference Network (HARN) and began a program of readjustment on a state by state basis. Wisconsin was the first such adjustment in 1989, and other state adjustments were performed from 1989 to 1997.
The center and orientation of NAD83 remained the same, but the NGS took the opportunity to bring the datum's scale into alignment with the ITRS (International Terrestrial Reference System) reference frame of 1989. The change of scale had negligible effect on the horizontal components of position, but it brought ellipsoidal heights in alignment with the ITRF89 frame, making GPS observation of height easier.
In 1994, the system of Continuously Operating Reference Stations (CORS) came into being, initially with about a dozen sites. The stations provided continuous GPS observation data and were positioned using the ITRF 1993 frame. The NGS adopted a 12 parameter Helmert transformation to convert between ITRF93 and NAD83, effectively defining NAD83 in terms of the ITRF 1993.
The transformation parameters were chosen so that the positions of 12 VLBI stations on the stable continental plate were transformed with minimal change to their NAD83 coordinates. The VLBI stations were selected because their locations were accurately known in both the NAD83 and ITRS datums, and VLBI geodesy provided the most accurate positioning technology, at an absolute uncertainty of less than 2 cm. The center and orientation of NAD83 remained the same, and the scale was made identical to ITRF93.
In 1996, the NGS performed a new conversion of all existing CORS stations, based on the 8 most stable VLBI stations. A new Helmert transformation using the ITRF 1994 realization was developed, again keeping the NAD83 coordinates of the VLBI stations as close to their original values as possible.
Again in 1998, the NGS determined a conversion for all existing CORS stations, again based on the 8 VLBI stations in the U.S. and adding 4 VLBI stations from Canada. A new Helmert transformation was determined using the ITRF 1996 realization. In 2002, the NGS recomputed the coordinates of all CORS sites using the ITRF realization of 2000 and supplied a Helmert transformation from ITRF2000 using an epoch of 2002.00 (i.e. the station's position on January 1, 2002). GPS accuracy is such that tectonic plate movement must be accounted for using velocity components, and an epoch is stated so that a station's current position can be computed based on its movement since the time of publication.
The adjustments that had been performed state by state beginning with the NAD83 (HARN) had resulted in small discrepancies across state boundaries. In 2005, the NGS began a new nationwide adjustment for all GPS stations, and scheduled the adjustment for completion on February 10, 2007, the 200th anniversary of the CGS. The adjustment included all GPS stations, did not include the older "classical" stations, and utilized the CORS stations as fixed control points, since the positions of the CORS stations are known to very high accuracy, safely allowing the assumption that there is no error in their position.
The new adjustment eliminated the discrepancies between states and provided published estimates of accuracy in absolute terms, e.g. +/- 5 cm, etc. In addition, both NAD83 and ITRF coordinates were published as a convenience for GPS users.
1. In 1845 the chronometric longitude of the Harvard observatory was accepted to be 4h 44m 31s.95 (W 71° 07' 59".25).
In 1851 Superintendent Walker adjusted the longitude to 29.5 seconds of time (W 71° 07' 22".50).
In 1867 the longitude was determined telegraphically to be 4h 44m 30s.85 (W 71° 07' 42".75)
In 1874 the "final" report of telegraphic longitude was 4h 44m 30s.98 (W 71° 07' 44".70)
In 1880 the least squares adjusted longitude was 4h 44m 30s.994 (W 71° 07' 44".910)
In 1897 the adjusted longitude was 4h 44m 31s.046 (W 71° 07' 45".690)
The NAD27 longitude was 4h 44m 30s.928 (W 71° 07' 43".920)
The NAD83 longitude was 4h 44m 30s.802 (W 71° 07' 42".030)
2. The longitude of San Francisco's Washington Square was first telegraphically determined in 1869 to be 3h 25m 7s.335 (51° 15' 50".025) west of Cambridge, Massachusetts. The accepted longitude of Cambridge at the time was W 4h 44m 30s.95 (W 71° 07' 44".250) making the longitude of Washington Square W 8h 09m 38s.325 (W 122° 24' 34".875). -- 1870 CGS Annual Report, Appendix 12