The reference system of the Canadian Geodetic Vertical Datum 1928 (CGVD28) is the Mean Water Level (MWL) as measured in ~1928 at five tide gauges: three on the east coast located in Pointe-au-Père, Halifax and Yarmouth and two on the west coast located in Vancouver and Prince Rupert.  CGVD28 does not represent an equipotential (level) surface because the mean water levels at each tide gauge are not at a same elevation.   Figure 1.1 [JPEG, 133.5 kb, 1264 X 940, notice] illustrates CGVD28 with respects to equipotential surfaces and mean sea levels.

For the new vertical datum, the reference system is the equipotential (level) surface (W₀) that coincides closely with the mean water level of the St-Lawrence River as observed over the last nineteen (19) years at the tide gauge (#2985) in Rimouski, Québec.  Basically, it is the same reference system as the North American Vertical Datum 1988 (NAVD88), which was adopted by the United States of America (USA) in 1995 as their reference surface for topographic elevations.  However, the realizations of NAVD88 and the new vertical datum for Canada are done differently.  NAVD88 is realized using spirit levelling observations while the new datum for Canada will be realized by modelling the equipotential surface using gravity measurements (geoid modelling).  Even though the two methods are independent, the two vertical datums would correspond to an identical surface if the realizations would be errorless.

Figure 2.1 [JPEG, 194.5 kb, 1142 X 850, notice] shows the difference between NAVD88 and CGG2005 in Canada.  CGG2005 is the latest gravimetric geoid model for Canada.  Its long wavelength components (longer than ~450 km) are determined using measurements from the GRACE satellites gravity mission.  The estimated geoid errors of the GRACE data are better than five cm.  Thus, the significant tilt observed from east to west is due principally to systematic errors in the levelling observations.

The equipotential surface for the new vertical datum in Canada is determined initially from tidal information at the gauges in Rimouski (#2985) and Pointe-au-Père (#2980), and from GPS and levelling observations at eight stations in the vicinity.  Figure 2.2 [JPEG, 106.9 kb, 1263 X 815, notice] shows the location of the tide gauges and control stations. The tide gauge in Pointe-au-Père was in operation from 1897 to 1983 before being transferred to Rimouski.  The gauge in Rimouski is active since 1984.    A first epoch GPS and levelling surveys were conducted in October 2006. 

The tidal information is available from the Marine Environmental Data Services (MEDS), Fisheries and Oceans Canada.  These data allows the determination of the mean water level (MWL) above the chart datum (CD).  This separation between the chart datum and the mean water level is represented by Z₀.  Its mean separation is estimated by averaging the last nineteen (19) years of monthly Z₀ values.  This information is summarized in Table 1.

Furthermore, the elevations of the reference markers at the tide gauges are also known with respect to the chart datum (H_CD).  This information is also available from MEDS.  Thus, it is possible to determine heights of the reference markers above the mean water level (H_MWL):

H_MWL = H_CD – Z₀.

The height above the mean water level (H_MWL) for the reference markers are shown in Table 2. Stations 9929000, 78K066 and 70L600 are measured with respect to tide gauge #2980 in Pointe-au-Père while stations 7629325, 9422901 and 94L4060 are measured with respect to tide gauge #2985 in Rimouski.

GPS and levelling measurements were conducted to four of these reference stations.  The other four stations observed by GPS and levelling are benchmarks (vertical control stations).   The stations in Table 2 ending with an "X" are temporary eccentric markers to the benchmarks with the same identification (without the "X").  These benchmarks could not be observed directly by GPS measurements.  The ellipsoidal heights at the eight stations were processed accurately in ITRF00 reference frame.  The accuracy of the ellipsoidal heights (σ_h) is approximately 3 mm.  The first-order levelling data between the eight stations was constrained to the geopotential number at station 94L4060, which is located next to tide gauge #2985 in Rimouski.  Its geopotential number (C) was determined using its height above mean water level:

C_94L4060 = H_MWL· g = 4.0569 m² s^-2,

where g is the mean gravity measured along the plumb line from the geoid, i.e., the mean water level and the Earth’s surface.  The heights from this local adjustment are identified by H₂₀₀₆.  The relative precision of the orthometric heights (σ_ΔH) is better than one cm.  The ellipsoidal heights and orthometric heights of these stations are summarized in Table 2.  The table also includes the published elevations for CGVD28 and NAVD88 as a reference.

Finally, the geoid heights (N) at the eight stations are determined from the GPS ellipsoidal heights (h_ITRF00) and levelled orthometric heights (H₂₀₀₆) as follows:

N_{GPS/Levelling} = h_ITRF00 – H₂₀₀₆

These geoid heights are compared to those determined from the geoid model.  In this case, as an example, we are using the CGG2005 geoid model while waiting for the model that will include data from the GOCE gravity satellite mission (to be launched in September 2007).  The mean discrepancy (ε) between the two sets of geoid heights is the bias to apply to the CGG2005 geoid heights in the Rimouski area to represent the actual geoid heights for the new datum: 

ε_i = N_{GPS/Levelling} – N_CGG2005.

The mean discrepancies will also allow the estimation of the potential value of the vertical datum:

W_Rimouski = W_CGG2005 – ε_i·γ_φ

where W_Rimouski is the potential of the mean water level at the tide gauges in Rimouski and Pointe-au-Père, γ_φ is the normal gravity on the ellipsoid at geodetic latitude φ, and W_CGG2005 (62636856.88 ± 0.5 m² s^-2) is the potential used to compute CGG20005 geoid model .  The estimated potential of the mean water level in Rimouski from the CGG2005 geoid model is shown in Table 3.  This mean geopotential value does not define the vertical datum.  The vertical datum is defined by the mean water level at Rimouski over the last nineteen years [e.g., 1987-2006].  The potential value of the geoid may change with each realization of the vertical datum due to errors in estimating the potential of the global mean sea level and to errors in the realization of the geoid model for Canada.

Station 9929000 [JPEG, 1.2 Mb, 2592 X 1944, notice], which is the CBN pillar in Pointe-au-Père, will become the fiducial station for the Canadian vertical reference system.  Its stability will be confirmed locally by levelling observations to the nearby benchmarks and regionally by GPS observations to the surrounding CBN stations.  Station 9929000 will be upgraded to an active control site, i.e., the station will be equipped with a permanent GPS receiver, which will track GPS signals continuously.  This will allow continuous monitoring of the Canadian vertical reference system.

Station 9929000 should be very stable.  The station is located near the hinge line (the zero isometric contour line) of the post-glacial rebound.  Thus, the ellipsoidal height of the station should be fairly constant over the years.  Figure 2.3 [JPEG, 241.9 kb, 2200 X 1699, notice] show a post-glacial rebound model for North America.