Vertical Coordinate Refresh

Last updated on January 17, 2022

Powell River Vertical Refresh

January 17, 2022

1.  Introduction

The Powell River ISA was vertically refreshed on January 4th, 2022. Prior to that date, the ISA was vertically referenced to a local vertical datum. This local datum was approximately 3.1 metres different from the newly established GNSS-derived CGVD28 datum in the municipality. The shift between the two datums varies throughout the ISA. Conversion between the two could not be handled via a simple shift correction for the entire ISA.

Click here to see the Local Datum to CGVD28 Transformation drawing illustrating the variable shifts throughout the ISA between the local datum and CGVD28.

In the era of ubiquitous GNSS use, surveyors in the region have long been arriving at their own independently derived CGVD28 heights using GNSS in conjunction with the HTv2.0 model. The large difference between the local datum and CGVD28 has been a source of confusion throughout the surveying and engineering communities. The alignment of the ISA’s geodetic infrastructure with GNSS-derived heights is therefore a big step towards elimination of these issues.

2.  A GNSS-Derived Realization of CGVD28

It is worth noting that CGVD28 is a very old datum that was defined well before the advent of GNSS technology. It was realized at the national scale via a high precision levelling network that was constrained to a select group of tidal stations on Canada’s coasts. It continues to be the official datum in BC as of the published date of this bulletin. Powell River is an isolated municipality and therefore does not have a direct high precision connection to this national levelling network. Proper realization of CGVD28 has therefore remained an impossible endeavor for this ISA. Doing so would require high precision spirit levelling along non-existent highways to the nearest federal benchmarks. That being the case, GeoBC has instead derived a pragmatic realization of CGVD28 via GNSS observations in conjunction with the HTv2.0 (Epoch 1997.0)​ model. The new published heights are best described as pseudo-CGVD28 heights due to their non-standard means of derivation. Elsewhere along highways and municipalities throughout BC, where federal benchmarks are abundant, this would be considered an indirect realization of the datum potentially conflicting with the established national levelling network of geodetic infrastructure. In Powell River, there are clear benefits to the pragmatic GNSS-derived realization. The -3.1m vertical step between projects referenced to monuments and projects referenced via independent GNSS observation is now eliminated. Aligning the ISA with CGVD28 (HTv2.0 1997.0) is a critical step towards the eventual eradication of the problematic local datum that has existed in the region for decades.

3.  Gradient Between Local Datum and New CGVD28 Heights

As noted in the Introduction, the difference between the 2 vertical datums is NOT a constant shift throughout the ISA. As can be seen in the Local Datum to CGVD28 Transformation drawing, the transformation between datums is variable throughout the ISA.

In the north end of the ISA, the shift is -3.09m meanwhile at the south end of the ISA the shift is -3.145m. A general transition in shifts (with some intermediate undulation) from north-to-south can be seen on the drawing. This amounts to a difference of 0.055m from end-to-end of the ISA. The shortest route distance following city roads from end-to-end of the ISA is approximately 13.6 km. This is the distance that would have been traversed during the original levelling campaign. There is therefore an average gradient of 0.004m per kilometre (following shortest route traversed during historic levelling) between the two datums. This gradient will fluctuate throughout the municipality and therefore surveyors are advised to review the Local Datum to CGVD28 Transformation drawing during new vertically sensitive projects. In such scenarios, surveyors are advised to assess whether the gradient between datums could cause any issues. The reason behind the gradient’s existence is difficult to prove, however it was not a surprise given the differences in survey methods used to realize each of the two datums. The local datum was derived from levelling observations, which are prone to systematic error growth over long traverse distances. A large number of instrument setups would have been required to derive a height difference from one end of the ISA to the other. Meanwhile, the pseudo-CGVD28 datum was established via GNSS observations which are directly affected by the relative accuracy of the HTv2.0 model. The new GNSS dataset will also have some relative vertical error present within. The 0.055m difference (over 13.6km) should therefore be thought of as a combination of the following:

  • small cumulative errors in historic levelling
  • small relative errors within the HTv2.0 model
  • small relative errors in the new GNSS observations

The HTv2.0 model is a hybrid geoid model that is not purely based on gravimetric data. However, it was compared with the CGG2013a model throughout Powell River at each GCM location. CGG2013a is a purely gravimetric geoid model. It is not errorless, but it was the best tool available to test for the presence of an artificial gradient within the HTv2.0 model. The two models where found to be close to parallel with one another throughout the ISA. The difference between geoid-ellipsoid separations (HTv2.0 – CGG2013a) ranges between 0.109m and 0.122m throughout the municipality; i.e. the largest relative height difference between HTv2.0 and CGG2013a within the ISA would therefore be 0.013m. Within the confines of the Powell River ISA, the HTv2.0 model is therefore very close to being parallel with a purely gravimetric geoid model (CGG2013a). The HTv2.0 model alone cannot account for the full 0.055m height difference over the ISA’s 13.6km route distance and is arguably a better vertical realization (with respect to gravity) than the previous local datum. Based on the limited information available, adoption of a GNSS-derived pseudo-CGVD28 datum is most likely partially rectifying an artificial gradient (with respect to gravity) that existed within the previous local datum. The vertical gradient between datums is therefore not expected to cause any substantial vertical issues. For vertically sensitive projects in Powell River, surveyors are advised to conduct high precision levelling between GCMs being referenced to verify that there aren’t any substantial issues being introduced by the newly published values. With the exception of outliers, listed in Table 1, relative height differences between adjacent monuments will be very similar to the previous height differences in the local datum. In the case of outlier GCMs, an improvement in relative height differences (over previously published values) is expected. Designation as an outlier implies a poor fit of the previously published values with neighboring GCMs.

4.  Method of Deriving New Heights

A total of 47 GCMs were directly observed via GNSS during the 2018 campaign. 11 of the 47 were observed via static GNSS meanwhile the remaining 36 were observed via RTK GNSS. The network of static baselines was constrained to Federal and provincial ACPs and was also confirmed via multiple CSRS-PPP solutions of the same datasets. The RTK GNSS survey was constrained to each of the 11 static GNSS stations throughout the ISA; static stations were used as RTK base stations. Of those 47 GCMs directly observed, 12 of them were found to be vertical outliers and are listed in Table 1 below. The remaining 35 directly observed non-outlier GCMS (observed static and RTK) were used as the basis for deriving heights for the remaining 170 GCMs not directly occupied.

The 170 GCMs not occupied had heights derived from an Empirical Bayesian Kriging height shift model. The model was designed to align all non-observed points with the 35 sample points that served as model constraints.  Each of the 35 input constraints provided known shifts between the datums. The modelled shifts would gradually transition between each of the constraints in an attempt to preserve (as much as possible) the previous (levelling-derived) relative height differences between adjacent monuments.

At outlier GCMs, changes in published relative height differences with their neighbors have undergone substantial changes. These changes are expected to yield an improvement to the affected relative height differences. As per standard practice, when discrepancies with published height differences are discovered, these should be reported to GeoBC via Marker Condition Reports.

5.  Conversion of Archive Projects from the Local Datum to CGVD28

Handlers of archive datasets must be cautious when attempting to convert old projects from the local datum to CGVD28. Any such conversions should be derived based on the published vertical shifts between datums at nearby monuments with some important exceptions noted below. This information is available on MASCOT long forms. The exceptions: monuments observed during the 2018 campaign that were flagged as vertical outliers when deriving the height transformation model. Outliers showed datum shifts that did not agree well with shifts observed at neighboring sample points. There are various reasons why a monument could be an outlier: disturbances, subsidence, upheave or errors in observations (old and new). These misfit monuments did not contribute to the height transformation model because their observed shifts are not likely representative of the true shift between datums at their location. For this reason, shifts published at outlier monuments should NOT be used for any conversions of archive datasets. Table 1 below shows the discrepancies that published shifts have with the estimated (model-derived) shift at these locations. In these instances, the model-derived shift can be assumed to be the more accurate of the two.

 
GCM Original Survey Method New Survey Method Change in Published Height (m) Estimated (Model-Derived) Difference Between Datums at this Location (m) Discrepancy (m)
229146 Static GNSS Static GNSS -0.12 -3.14 3.02
272575 Levelling RTK -3.25 -3.12 -0.13
677088 Levelling RTK -3.08 -3.14 0.06
222448 Levelling RTK -3.08 -3.14 0.06
79889 Levelling RTK -3.11 -3.14 0.04
307850 Levelling RTK -3.17 -3.14 -0.03
383984 Levelling RTK -3.08 -3.11 0.03
443341 Levelling RTK -3.07 -3.10 0.03
676999 Levelling RTK -3.11 -3.13 0.02
625913 Levelling RTK -3.10 -3.12 0.02
422956 Levelling RTK -3.16 -3.14 -0.02
643296 Levelling RTK -3.13 -3.14 0.01

Table 1: Vertical Outliers Among Sample Points

 

The error budget of a given dataset must be taken into consideration before converting in the manner described. As well, it should be assessed whether datasets were constrained to outlier monuments in the past – if that was the case, then the appropriate conversion would be different than what is described here.

The manner in which a dataset is converted should be clearly marked on any plans or documents. For high precision applications, consider resurveying archive project control relative to nearby GCMs. A resurvey will be the most reliable means of integration with the new CGVD28 realization in the ISA. As per routine standard procedure, surveyors are advised to occupy multiple GCMs and compare their observations with published values in order to detect erroneous published values. This is especially important for the outliers listed in Table 1. As mentioned above, it is expected that surveyors will now see better agreement in relative height differences between outliers and their adjacent monuments. However, this assumption has not been field tested. If the new relative height differences between outliers and their neighbors are found to be poor, this should be reported to GeoBC for further review. After having considered the cautions above, click here for a listing of the dN, dE, dH changes applied to each monument in the ISA.

6.  Potential for Negative Elevations on Future Surveying Projects

All new published heights of GCMs continue to be greater than 0m above the datum. The lowest monument in the ISA is GCM 643296, which has a CGVD28 height of 3.352m. The CGVD28 heights of topographic features in the “general shoreline area” in Powell River now will be approximately 3.1m lower than their previous heights relative to the local datum. Side note: use of the term “general shoreline area” in this bulletin is being used vaguely describe the swath of dry land, close to the water’s edge that has the lowest elevations in Powell River. In the context of this bulletin, the term is NOT specifically referring to the location of the natural boundary or chart datum.

At this point in time, the existence or extend of lands in Powell River that fall below the 0m elevation mark within the CGVD28 datum have yet to be reviewed. There is therefore a potential for portions of future shoreline projects to have negative elevations relative to the CGVD28 datum. This is a potential drawback to the establishment of CGVD28 in Powell River which may not have existed during the local datum era. For projects close to the general shoreline area, surveyors will now need to exercise care to mitigate errors related to a mix of positive and negative height values of topographic features within the same dataset. Pending receipt of further information, this aspect of the bulletin can be revised at a future date to clarify the circumstances.

7.  Previous Datum Labelling on MASCOT During the Local Datum Era

Prior to January 4th, 2022, long forms for GCMs in Powell River were stamped with “CVD28BC” as the vertical datum and “NOT INTEGRATED” as the integration status. These labels have likely been posted since the initial publishing of values for Powell River GCMs in the 1990s. Due to the incorrect datum label on MASCOT long forms (spanning approximately 25 years), there are likely many archive survey plans and datasets that incorrectly have CGVD28 recorded as the vertical datum. If these plans and datasets had heights that were derived relative to ISA monuments, then they were referenced to the local datum and are not compatible with the latest CGVD28 (HTv2.0 1997.0) realization. Archive datasets therefore require careful review to determine the source of the (alleged) CGVD28 heights. In many cases, a resurvey of project control might be necessary to avoid -3.1m errors based on false assumptions.

8.  GCMs with Old GNSS-derived Positions From 1990s

Prior to January 4th, 2022, there were 3 GCMs in/nearby the ISA that already had GNSS-derived heights from old observations made in the 1990s: 229146, 301424 and 726661. These three GCMs were incompatible with the local datum during its existence. Fortunately, GCM 229146 and 301424 were directly observed in 2018. Their previously published heights were relatively close (by comparison to all other ISA monuments) to the newly established CGVD28 datum: they required relatively small decimetre-level corrections (-0.12m and -0.19m). As a result, they are now compatible with the remaining 200+ monuments. Unfortunately, GCM 726661 was not directly occupied in 2018. Deriving a CGVD28 height for this individual monument was not possible via the bulk modelling process. The accuracy of the old published value (relative to the newly established CGVD28 datum) is unknown and is therefore a potential source of error and confusion for new surveys. In order to mitigate issues related to the incompatible height, this GCM has been pulled from public circulation and can no longer be found on MASCOT.

9.  The Long-Range Future of Height Referencing in Powell River

On the horizon, GeoBC intends to eventually adopt CGVD2013 province wide. Powell River is among a small group of municipalities in BC where the historic terrestrial network has a substantial amount of high quality GNSS-sampling. A beneficial by-product of this current initiative is that the Powell River ISA can be readily (and accurately) converted to the CGVD2013 datum. Unlike CGVD28, realization of the CGVD2013 datum does not require a high precision levelling connection to the network of federal benchmarks.  For many municipalities throughout BC, this is currently not the case. The ISA is therefore considered to be in a favorable state with respect to the future vertical datum.