In previous issues, we have briefly visited three of the “seven C’s” of NOAA’s National Geodetic Survey (NGS), as best methods for real-time (RT) GNSS positioning. We can now move to the fourth “C”--constraints--to examine whether we should constrain coordinate solutions to the passive control marks in the ground or to active control points, such as those from a real-time network (RTN), or both. First, let’s consider our current control situation. There are approximately 107 RTNs operating in the United States providing coordinates to users from “active” control stations, i.e., those that are accessible remotely 24 hours per day, seven days per week, 365 days per year. NGS is encouraging these RTNs to be aligned with the National Spatial Reference System (NSRS) as realized through the CORS network at 2 centimeters horizontally and 4 centimeters vertically (ellipsoid height). Additionally, the NGS hybrid geoid model enables users to obtain NAVD 88 orthometric heights from these values (see Figure 1) but at variable accuracy across the country, primarily because of differing density of ellipsoid height observations on bench marks and the availability of good gravity data.
Users who obtain positions from RTNs, CORS or OPUS will find that they do not always match the coordinates of the passive control to within the needed accuracies for their project. While there are more than 1 million passive marks (think concrete post in the ground) in the NGS database, many may have been disturbed or subjected to local or regional movement or have even been destroyed. Unlike an active station, the last thing we know about a passive mark’s position is where it was when it was last observed and published to the NGS database. When we walk away from such a mark, it’s the last time we know about its true position.
Some bench mark monuments in the database have not been geodetically leveled in more than 60 years. Since everything is moving, there is the possibility for uplift, subsidence or other movement over that time--not to mention acute movement, such as earthquakes. At just 1 millimeter per year, a bench mark could easily have a several-centimeter bias relative to its NAVD 88 height as published on the data sheet. Therefore, these passive marks’ coordinates may vary from those obtained from the RTNs or from GPS surveys tied to the CORS network. Furthermore, the RTN’s base station coordinates may also vary from the CORS coordinates by several centimeters as well. Add into the mix the fact that we have differing field methods and differing field conditions for RT, different adjustments and epoch dates of NAD 83 that RTNs use, and the different campaigns and methods used for the passive marks’ published coordinates, and it’s no wonder we get so many different answers for our positions!
So what do we do about this situation? Do we ignore passive marks? Well, of course the answer is–it depends. We usually find that RTNs can give us acceptable precision for our project work in the horizontal components if we are working in the same datum upon which the RTN is based. Usually we can tie projects together to within a couple of centimeters. However, in many cases (with some notable exceptions), the orthometric heights or “elevations” we produce using the RTNs derived ellipsoid heights with the NGS hybrid geoid model do not give us acceptable accuracy. This puts us at the crossroads where we have to ask whether the existing passive control in a project area is integral to the work produced or producible. Has it been used to produce design work, historical plans or other project data such as LiDAR, photogrammetric plans or digital imagery? Will it be used by us or by others for project work in the future?
If we are staking out structures, infrastructure, utilities, dwellings, etc., that have been recently designed from existing local passive marks, we must hold those values if they are represented in site monumentation. Do the RTN-derived coordinates vary, either vertically or horizontally, from the passive marks’ values more than our required project accuracy? Is the precision of how the passive control actually fits together unknown? If the answer to any of these questions is “yes,” then we should consider constraining the passive control by performing a least squares (best fit) adjustment. This essentially creates a transformation using the field GNSS obtained values to the local coordinates and provides an inclined best-fit project plane across the passive control. The actual alignment of the passive control to the NSRS is completely disregarded in this case.
The NGS hybrid geoid model can be used within this “calibration” or “localization” to provide better heights than with the inclined plane alone. Routines in the various GNSS manufacturers’ field data collectors or office software may vary, but they are generally performing the same process. As a result, though they are typically untraceable geodetically back to the datum, they provide an easy, proven method to lock the local control values together to a site. Because of the uniqueness of these calibrations, the metadata and control used should be published or maintained with the project data.
The next RT article will deal with the fifth “C”--coordinates. This will include a discussion of precision versus accuracy and how differing sources provide different values for our RT work.
Recommended Criteria for a Constrained Site• Use at least four control points that surround the project site and form a rectangle to the best extent possible. This geometry helps prevent errors from occurring, because an outlier will be apparent. These marks should provide trusted values for what is being adjusted--the vertical and/or horizontal. The more trusted marks, the better, especially on site.
• An autonomous point can be occupied on site to perform the calibration or site control can be established after the constrained adjustment is performed.
• The adjustment can be performed in the field or in the office.
• Make sure the field party and the office are using the same adjustment. Serious errors can occur if this is not done.
• Do not perform work outside the adjustment envelope. Perform an expanded adjustment to further marks if necessary. It should be understood that any new calibration will change existing coordinate values to some extent.
• The constrained adjustment is a great way to see how the control fits together. We assume good field GNSS practices are followed, which means our ellipsoid heights and projected coordinates will be precise. Reports can be generated in the software that show the maximum horizontal and vertical error, the individual residuals, the scale of the solution (hopefully near “1”) and the vertical slope in parts-per-million (ppm).
• Although not recommended, a two-point vertical localization can be done. If there are two trusted bench marks on site, one can be constrained and one used as a verification. All elevations would then be shifted to use the held value as truth. You must have confidence in this value and understand that everything hangs on its accuracy along with that of the geoid model used.
• RT GNSS can indeed produce satisfactory, relative elevation values with the same initialization over a small project area, using straight ellipsoid height differences and ignoring the hybrid geoid model.
• Constraining a project passive control envelope.
Review the First Three C's:1. Checking Your RT Gear
2. Communication in Real-Time Positioning
3. Conditions Influencing RT Positioning