NGS 2018 GPS on BMs program in support of NAPGD2022 — Part 8

My last two columns (NGS 2018 GPS on BMs program in support of NAPGD2022 — Part 6 and NGS 2018 GPS on BMs program in support of NAPGD2022 — Part 7) described the National Geodetic Survey’s (NGS) GPS on BMs 2018 interactive web map, and provided an update and status report on stations observed in support of the 2018 GPS on BMs Program. This column will provide another update and status report on stations observed in support of the 2018 GPS on BMs program and provide an example of how the OPUS-shared results filled in a void area in West Virginia that will benefit the development of the hybrid geoid model GEOID18. The column will also provide an example of how OPUS Shared results identified a reset station that has an invalid NAVD 88 height, and the importance of having a least two OPUS Shared results to ensure the reliability of the OPUS solutions.

As mentioned in the last column, the GPS on BMs 2018 web page contains a link to a web map where users can determine which bench marks NGS would like users to occupy before the August 31, 2018, deadline. The box titled “2018 Web Map” depicts the map update as of July 27, 2018 (1738 priority marks completed). My last column reported that as of May 29, 2018, there were 1067 priority marks considered completed. During the past two months, 671 more priority stations have been reported completed. This is progress but this still only represents about 30 percent of the priority marks. Hopefully, this will increase dramatically during the month of August. Remember, the cut-off date for data to be included in the creation of the hybrid geoid model GEOID18 is August 31, 2018.

NGS periodically provides an update on the GPS on Bench Marks Program. On July 3, 2018, NGS sent an email to everyone that shared GPS data on NGS bench marks via OPUS or registered for NGS’ February 2018 webinar about GPS on Bench Marks. The email provided an update on the GPS on Bench Marks Program (see box titled “July 3, 2018, NGS Email on GPS on BMs Update”). The map provided in the update indicated that some of the new observations may generate changes between +/- 8 cm.

NGS is tentatively planning another webinar on the GPS on Bench Marks program for August 9, 2018 (2 pm to 3 pm eastern time). NGS will provide an update on the GPS on Bench Mark program and probably will highlight potential improvements between the current hybrid geoid model GEOID12B and the latest prototype version of the future hybrid geoid model GEOID18. I would encourage everyone to sign up for the NGS webinar series.

Source: Plot Generated Using ArcGIS

As indicated in the figure provided in NGS’ July 3rd update on the GPS on Bench Marks program email, there are many areas of the country that have already benefitted from users participating in NGS’ GPS on BMs program. This column will highlight an area near Charleston, West Virginia, were users have been very active in providing OPUS Shared results. The box titled “GPS on Bench Marks near Charleston, West Virginia” depicts the marks that meet NGS’ criteria and will be involved in the development of the hybrid geoid model GEOID18. As you can see from the plot, there are several new stations that will be used in the development of the model which will help to improve the reliability of the product.

The box titled “An Example of OPUS Shared Stations in Charleston, West Virginia, Region” provides the stations’ PID and OPUS designation. The six OPUS Shared stations cover approximately a 50 km square area. Most of the stations are only 10 km apart. These stations will definitely help to improve the reliability of the hybrid GEOID18 model.

When using OPUS Shared results, users should always check to see if a station has been observed more than once. The box tilted “Differences in OPUS Shared Ellipsoid Heights in Charleston, WV, Region” lists the pairs of OPUS observations for the stations depicted in the previous plot. The column labeled “Difference in Ellipsoid Heights” provides the differences in ellipsoid heights based on the two different OPUS Shared results. All differences are less than 1.5 cm and most are less than 1.0 cm. This is indicating good repeatability to the cm level but this may not be indicating accuracy. These stations were observed one day apart but observed at about the same time of the day. They could have the same systematic errors effecting the results such as multipathing and satellite geometry. When performing the second OPUS Shared observation, users should select a different time of day to improve the chances of detecting, reducing, and/or eliminating the effects of remaining systematic errors.

The box titled “Differences in OPUS-Shared GNSS-Derived Orthometric Heights Using GEOID12B and Published NAVD 88 Heights” provides the differences between the GNSS-derived orthometric heights using GEOID12B and the published NAVD 88 values. This table indicates that there is a large difference (23.4 cm) for station HX2382 (L105 Reset 1962). Since the two ellipsoid heights only differ by 1.0 cm, this is an indication that the station probably moved since it was Reset or the reset observations were performed incorrectly. Either way, this station should not be used in the development of the hybrid model or used by anyone for geodetic control.

Since GEOID12B is a hybrid geoid model that was designed to be consistent with NAVD 88 values, I always compute differences between GNSS-derived orthometric heights using the experimental geoid model and published NAVD 88 height values. I described this process in my October 2015 column (https://www.gpsworld.com/establishing-orthometric-heights-using-gnss-part-3/). The box titled “Differences in OPUS-Shared GNSS-Derived Orthometric Heights Using xGeoid17b and Published NAVD 88 Heights” provides the differences between the GNSS-derived orthometric heights estimated using IGS08 (2005) ellipsoid heights with the xGeoid17b geoid model and published NAVD 88 heights. The values in the column labeled “GNSS-Derived Orthometric Height minus Published NAVD 88” represent an approximate difference between NAPGD2022 and NAVD 88. The box titled “OPUS-Shared GNSS-Derived Orthometric Heights Using xGeoid17b minus Published NAVD 88 Heights” provides a plot that depicts these differences.

Once again, it should be noted that PID HX2382 value is much different from the other values. To look for outliers, a mean difference was removed from the results. The box titled “OPUS-Shared GNSS-Derived Orthometric Heights Using xGeoid17b minus Published NAVD 88 Heights with a Mean Value Removed” makes it easier to see that station HX2382 is an outlier. The station is approximately 25 cm different from its neighboring stations that are only 10 km away. As previously mentioned, this station apparently moved since being Reset in 1962 or the reset observations were performed incorrectly. Identifying stations that have moved since the last time they have been leveled is one of the benefits of participating in the GPS on BMS program.

For completeness, both a bias and trend were removed from the differences since IGS08 (2005) GNSS-derived orthometric heights and NAVD 88 heights indicate that there’s an apparent long-wavelength trend between the two sets of values. The box titled “OPUS-Shared GNSS-Derived Orthometric Heights Using xGeoid17b minus Published NAVD 88 Heights with Bias and Trend Removed” depict the differences with a bias and trend removed. As in the other figures, PID HX2382 clearly indicates that it is an outlier relative to its neighbors. This station would be rejected by the geoid team when creating the next hybrid geoid model.

It should be noted that except for the Reset station, all of the differences are less than 2 cm. Although, some relative differences between closely-spaced stations approach 4 cm. For example, the differences between stations HX1746 and HX2496 is -3.7 cm (-1.8 cm – 1.9 cm). The differences in ellipsoid heights from the OPUS Shared solutions are all less than 1.5 cm, even the differences between ellipsoid heights for station HX2382 is only 1 cm. This is an indication that the reset station, HX2382, does not have a valid NAVD 88 published height and should not be used as control. Surveyors that adhere to the FGCS specifications and procedures would realize that this station did not have a valid NAVD 88 height and would not use the published NAVD 88 as control in their project. For example, surveyors performing a leveling project would perform a 2- or 3- mark leveling tie and the results would indicate that the station had moved since it was last leveled.

This type of validation procedure should also apply for OPUS users. If a user obtains one OPUS solution and proceeds to perform a survey from that station, the user does not know whether the OPUS height value is reliable or accurate. One solution does not provide any indication of reliability.

(Source: Merriam-Webster dictionary)

The OPUS Shared station PID SV0942 (A 25) is an example of two OPUS Shared results generating ellipsoid height values that differ by 10 cm. (See yellow highlighted section in the box titled “Differences in OPUS Shared Ellipsoid Heights for PID SV0942.”) This large difference is significant when you performing a survey where you need heights to better than 3 cm (0.1 foot). This is one reason that NGS requires two OPUS Shared solution for every mark used in the development of the hybrid geoid model.

In the OPUS Shared solutions of PID SV0942, the latest OPUS Shared GNSS-derived orthometric heights (2018-07-14) agrees to about a cm with the published NAVD 88 height, while the 2014 Opus Shared GNSS-derived orthometric height is -11.4 cm different from the published NAVD 88 value. (See yellow highlighted section in box titled “Differences in OPUS-Shared GNSS-Derived Orthometric Heights Using GEOID12B and Published NAVD 88 Heights for PID SV0942.”)

It should be noted that the error estimates provided in the Opus Shared output indicate the ellipsoid heights are good to about +/- 1 cm. (See highlighted section in box titled “Two OPUS Shared Solution for PID SV0942.”) Saying that, the two NAD 83 (2011) ellipsoid heights disagree with each other by 10.2 cm. I like a quote that is attributed to Mark Twain – “It ain’t what you don’t know that gets you into trouble. It’s what you know for sure that just ain’t so.” (Obtained from http://lukefostvedt.com/famous-quotes-about-statistics/). I’m not suggesting that Opus Shared solutions results are incorrect. I’m attempting to provide an example of why users need to repeat all observations and to demonstrate how error estimates can be misleading.

The number of GPS on Bench Mark stations completed as of July 27, 2018, represents about 30 percent of the total number of stations that need to be observed. As I have explained in previous columns, there are many invalid GPS on BMs stations that may be used in the next hybrid geoid model unless more bench marks with valid NAVD 88 heights are observed with GNSS. NGS will accept data for inclusion in the next hybrid geoid model, GEOID18, until the end of August 2018. After that, NGS’ GPS-on-Bench-Mark Program will expand to include other regions and will focus on data to improve NGS datum transformation tools. This column provided an update and status report on stations observed in support of the 2018 GPS on BMs program, provided an example of how the OPUS Shared results can be used to identify a station that may have moved since it was last leveled, and the importance of repeating OPUS observations. I would encourage users to register for NGS’ next webinar on the GPS on Bench Mark Program scheduled for Thursday, Aug. 9th to hear the latest status of the program.