San Francisco Precision Survey Completed
When Bruce Storrs first accepted his job as San Francisco City and County Surveyor 14 years ago, he was well aware of the challenges. “The first thing I thought was, ‘If there’s a seismic event, and I get a call from the Director of Public Works asking where things are now so we can rebuild,’ how am I going to answer that?”
His response turned out to be a new high precision-network (HPN), based on 20 monuments spread strategically throughout the City and County of San Francisco, together with a new low-distortion grid projection, CCSF-CS13, specially crafted for the region.
I Want It Done Right
The project really got started in 2012, when Michael R. McGee, PLS, head of Santa Barbara-based McGee Surveying Consulting, was recommended to Storrs as an expert in setting up GPS networks in cities. “That was great news,” Storrs says, “because I knew McGee. We went to school together at CSU Fresno, and it was great to rekindle our friendship. Fortunately, he was also the right choice for this project — I’ve never seen a network for a city that is more rigorous or better supported than what we have here.”
McGee had completed similar projects in Santa Barbara, Stockton, Santa Monica, and Los Angeles, and had also established larger scale control networks for 927 miles of California coastline and 2,500 square miles of China Lake Naval Weapons Center. But even with all McGee’s experience, the CCSF Geodetic Network was something of a magnum opus.
“Every project is different, depending on the resources and equipment available,” McGee explains. “I work with the client to establish just how precise a network they’re after, how much they want to spend, and how they want to maintain it. Then I come up with a project design that works.”
In the case of San Francisco, McGee’s brief was simple in concept. “I want it done right,” Storrs said, asking only for 20 monuments. (By contrast, McGee had established 130 control points in Santa Monica, and 400 in Los Angeles.)
But in light of their potential importance in the wake of an earthquake, the 20 points in San Francisco needed to be special. To accurately track movement of an entire major city-county, the points had to be especially precise. And, to be useful when they were needed most, they had to be both quickly and easily accessible. For McGee, it was a major professional challenge that would take years to complete.
For Storrs, it was more of a political and bureaucratic challenge. “I had the luxury of answering only to the director of public works here, and budget was available from a fund that had been building up since 2005 and from subdivision applications that support all the boundary surveying in the city,” he explains. “That was fortunate — explaining the benefits to the Board of Supervisors would have been difficult, because this is a highly technical project, and the payoff was not immediate. It helped that McGee was very frugal, and he was our only outside consultant on this — we did all the field work with our own staff. That meant we didn’t have to cut corners.”
McGee put standards for monumentation and field work in place that were exceptionally rigorous and paid attention to technical and human factors throughout the project. With the help of Bill Hurdle, a southern California surveyor, McGee even designed a custom projection for the project. The City & County of San Francisco Coordinate System (CCSF-CS13) is a low-distortion grid projection system designed for CCSF to be nearly a ground coordinate system. The City varies from sea level to approximately 1,000 feet in elevation. To minimize the distortion between ground and grid, the projection surface was positioned at the most common ground height in the County taken at an ellipsoid height of 44.50 meters (146.0 feet).
Meanwhile, Storrs supported him on the bureaucratic end, allowing McGee and city staff to do their best work and create an absolute jewel of a network and an amazing foundation for infrastructure work in the notoriously unstable Bay Area. “It was an honor to work with such committed public employees whose dedication and contribution to the success of this project made all the difference,” McGee says.
Best-Ever Municipal HPN?
It wasn’t as if San Francisco was starting from scratch. The area is densely packed with high order control points based on NAD83 and benchmarks based on NAVD88 as well as a previously used Old City Datum. Work on the new HPN began with recovery of many of these points by means of high precision leveling and a GNSS survey, performed with Leica Viva GS15 receivers. The NAD83(2011) Epoch 2010.00 datum and reference frame were recovered by relying on four CORS that bracket the county. The NAVD88 datum was recovered based on 35 NGS benchmarks.
In his survey report for the City & County of San Francisco 2013 Leveling Network, McGee notes, “This survey included a ‘Prime’ City benchmark monument known by the NGS as HT0781 which is a chiseled triangle in the top of a granite step located on the NW corner of Townsend Street and 2nd Street,” and, “The NAVD88 height of the ‘Prime’ City Benchmark determined in 2002 is identical to the height of 7.158 meters determined by the CCSF 2013 Leveling Survey.”
Thirty-five benchmarks eventually served as the basis for establishing a conversion factor from the previous city datum and as a basis for recovering the NAVD88 datum in the city.
Actual work on the HPN began in January 2013 with a high precision vertical control network based on differential leveling within the City and County of San Francisco. The work was a model of best practices for urban leveling work. “We did 72 miles of leveling, the most precise in all the years of survey work here,” says Storrs. All leveling was performed with a Leica DNA10 electronic digital level and a pair of Leica GKNL4 fiberglass bar code rods. The DNA10 was calibrated by Leica Geosystems prior to the survey and a level collimation test (peg test) was performed prior to each field day of operation.
From January to October, a three-person crew surveyed 22 loops that included all 20 points in the HPN. Average closure for the 22 loops was 3 mm (0.01’), and 20 of the loops closed within NGS First Order, Class I standards. One notable procedure was monthly calibration of the fiberglass rods, and rod seams (see Calibrating Fiberglass Rod Seams).
The GNSS survey work began in July 2013 based on 20 new monuments set in ideal locations that satisfied a rigorous set of criteria established by McGee (see Placing Precision Points). All GNSS work for the city network was done over five days featuring consistently temperate, overcast weather, using four Leica GS15 GNSS receivers. The city network contained 83 vectors averaging 4.3 km (2.7 mi.) in length. Each point was occupied four times for 45 minutes, with a minimum of six GPS satellites and six GLONASS satellites observed, and up to 21 total satellites observed. The 1D and 2D residuals averaged 0.003 meters (0.01 feet). The regional network contained 57 total vectors averaging 20 km (12 mi.) in length, and each vector represented three 24-hour observations staggered every other day. The 1D and 2D residuals averaged 0.002 meters. Since the Bay Area is crossed by multiple fault lines, McGee’s observation protocols were switched up from day to day during the five-day campaign to provide redundancy and to account for independent movement of the four CORS utilizing the NGS HTDP model.
After extensive processing, results were excellent, meeting the NGS specifications per the report at the 95 percent level of confidence as follows:
- Local Horizontal Accuracy Classification is 5 mm
- Local Ellipsoid Height Acc. Classification is 5 mm
- Network Horizontal Accuracy Classification is 1 cm
- Network Ellipsoid Height Acc. Classification is 1 cm
Throughout the project, McGee paid attention to the human factor. For example, not only were project-specific observation and daily work protocols established, they were practiced in prior training, and all field staff were tested to demonstrate their knowledge and competence in the procedures.
Publication, Verification and Acceptance
The majority of this network design and fieldwork was completed in 2013, and the results were published on a recorded map (along with the new projection, CCSF-CS13) in early 2014. The new high-precision network is now officially known as the City & County of San Francisco 2013 High Precision Network, or CCSF-HPN.
In March 2018, the CCSF-HPN was independently resurveyed and extended with the same care and effort that went into the original survey. “We reproduced the original survey fieldwork with different receivers and personnel, occupying all the original points and establishing eight new control points,” says McGee. “Basically, we completed an independent re-survey of the entire network.”
Results were remarkable; precision of the new survey was so good that original results were confirmed at the level of 3 millimeters on average across the network. Except for the southwest corner of the city (nearest to the San Andreas Fault), the original points were in the same relationship to each other. “In other words, you could drop the 2018 network onto the 2013 network and the points match 1-6 millimeters (0.00-0.02 feet) 2D,” McGee explains. “We determined the network had moved N 33˚ W 15.5 centimeters (0.51 feet) — that’s how much San Francisco City-County moves in 4.64 years, or 3.3 centimeters (0.11 feet) per year. What’s important is that it moved as a block, no rotation or stretching.” It’s believed to be the most precise measurement ever obtained of a California county’s seismic movements.
Since 2013, the CCSF-HPN has become widely used in the city and county, and CCSF-CS13 has been adopted and made available on major platforms like ESRI and Autodesk. “It’s taken on a life of its own,” says Storrs. “Every major new subdivision — that’s 60 or so now — is tied in. And because we do the field work for most other agencies, like public utilities and transportation authorities, we’re also extending this high precision network to many more monuments throughout the city — a couple hundred already, and up to several thousand on our six to seven major thoroughfares.”
This new precision has also been extended to San Francisco County boundaries; using the new Leica GS18T GNSS RTN. The City staff resurveyed the six-mile San Francisco-San Mateo County Line much more efficiently and precisely than any previous resurvey. In fact, the county line had not been re-surveyed since 1898.
Designed and surveyed with extraordinary care and skill, the City and County of San Francisco’s Geodetic Network may well be the most sophisticated and precise regional high-precision network yet created. And it’s doing far more than simply serving as a baseline for rebuilding after a seismic event or even the “Big One” that every Bay Area resident dreads.
“It’s revolutionized infrastructure in the city,” Storrs explains. “It’s become the coordinate system that unifies public safety, utilities, police, parcel mapping, you name it. Everything is coming together on this phenomenally precise network, and there’s very little grid-to-ground difference (0.01 feet per 1000 feet) in the new projection, so we can survey across the city with virtually no adjustments. It’s so much more than I anticipated.”
To learn more about the CCSF Geodetic Network, visit http://sfpublicworks.org/ccsf-geodetic-network. To learn more about the surveying solutions used in this project, visit https://pure-surveying.com/.
In the Summer 2013 issue of California Surveyor, Michael McGee, PLS, and Robert Reese, PLS, published their method for calibrating fiberglass rod seams, along with a calibration form that was used by McGee and field staff during the high precision leveling runs for the CCSF Geodetic Network. The article, Seam Errors on a Digital Bar Code Level Rod, asked, “So you got your new bar code digital level to do that first order work your new client requested. Nice. And you bought the three section rods, understandably, because the one-piece invar rods are so darned expensive and a little difficult to carry around. How good is the rod reading value at the top? Is it really 4 meters? Or is it 3.999 meters? Or worse, you really don’t know?”
The recommended three-reading calibration method is straightforward:
- Set up an area (concrete steps make a pretty good test site) with stable, well-defined benchmarks, approximately 2-3 feet difference in elevation.
- Set up the level so it sees the bottom section of the rod on both benchmarks.
- First reading: Read the rod’s bottom section on Benchmark “A.” Read the rod’s bottom section on Benchmark “B.” The difference between the rod readings is the apparent true difference between “A” and “B” (H1).
- Second and third readings: For the second reading, set up the level so it sees the rod’s middle section on “A” and bottom section on “B.” Likewise, for the third reading, set up the level so it sees Rod’s top section on “A” and mid section on “B.” The difference between readings H1, H2, and H3 can be used to calculate rod seam errors between bottom, middle, and top sections. Accounting for these errors avoids accumulated inaccuracies in long level runs.
Notably, this procedure and form was used successfully on the city-wide leveling project in the City and County of San Francisco.
The actual location of high precision points is obviously a critical factor, especially in dense urban areas. Since all 20 points in the new CCSG Geodetic Network needed to be precisely survey-able, stable and recoverable long term, and easily accessible for routine work in the wake of a seismic event, demanding guidelines for point placement were established. Criteria for GNSS site suitability included:
- Clear horizon above 15 degrees
- Avoid sources of multipath like buildings (50’ minimum), water surfaces and signs
- Avoid nearby radio frequency sources and microwave paths
- Unlikely to be disturbed for many years
- Safe during occupations, with adjacent parking
- 24-hour unhindered accessibility
- Sited on stable geology not situated on a side hill or otherwise on large concrete structures