Followers 0

## Recommended Posts

I was kind of bummed on my walk today thinking about how NGS support for the database of survey marks is going away soon. I love these objects, love searching for them, love the whole concept. What to do? It dawned on me: why not make my own horizontal control mark? Right in my own backyard! So here's my plan. I wonder if anybody else has done this, and if so, can they offer any advice?

First step would be to purchase an appropriate brass disk from Berndtsen. Then with a number of bags of redi-mix and archived instructions from CGS I should be able to pour a "genuine" monument and mount a disk pretty easily. The hard part, of course, would be to measure the position of the disk with some kind of reasonable accuracy. As an amateur without surveying equipment (or experience!) I would have to rely on GPS. As everybody knows, a single GPS observation ( obtained, say, from an iphone) is not very accurate. The CEP (circular error probable) is about 10 meters, while bona fide horizontal controls have a CEP of a couple of centimeters. If you took multiple independent GPS observations, however, and averaged the results, statistics comes into play. With N independent observations the CEP gets divided by the square root of N. The square root of a million is a thousand, so with million observations averaged I'd get about 4 inch accuracy. I could live with that.

How to get a million observations?  I own a Garmin Montana GPS that can record gpx files. I'd lay that on top of my brass disk set to record a point every 5 seconds. If I let it run for maybe two hours every day I'd have enough data in a couple of years. (Of course, there are mathematical issues to deal with. E.g., observations taken 5 seconds apart would not be independent, but with so much data it should be possible to measure the correlation coefficient and adjust for it appropriately. More observations needed, probably.)

Ah, yes, nothing new under the sun. I should have known.

I do have have a couple of related questions, though. I've been assuming that with sufficiently many independent GPS observations averaged, you can get as close as desired to the true location. This seems not to be true, though. According to GPS.gov the limiting accuracy of N averaged observations as N goes to infinity is about 2 feet. (Not nearly as accurate as adjusted horizontal control.) But I seem to remember hearing that professional grade gps recorders can get accuracy close to the survey mark accuracy by observing for 4 hours or so. If so, how do they do it? Do they use more satellites? (US + GLONAST + Chinese?) Note: the GPS accuracy of +/- 10 meters that people throw around is a different thing. It includes errors introduced by the receiver, rather than the accuracy of the source.

Second, if an amateur, having mounted his/her own disk, can establish that their method is sufficiently accurate, could they apply for and get a PID? Has anybody ever done it?

I did some extensive averaging experiments with my Garmin a decade ago, and found that there were computational errors (roundoff of intermediate values?) that limited the resolution to a foot or two. Averaging might still reduce the error below that.  I was getting differences of several feet from day to day on 2-hour averages, and did a lot of days, but never felt I was getting to the foot level.

Professional grade receivers get their accuracy by using carrier phase rather than the code timing that most handhelds use, receiving multiple signals from the satellites (GPS L1 and L2, for instance), sometimes receiving multiple constellations (GPS, GLONASS, GALILEO, BEIDOU), correcting for iono/tropo delays by comparison to known stations which are relatively nearby that received signals from the same satellites at the same time, and using time averages. The known stations are either

a) fixed CORS stations or proprietary network stations run by Trimble, Leica, etc.  This processing is done by NGS program OPUS or vendor proprietary software. The fixed stations are typically a few 10's of km away and multiple station comparisons are made.

Or b) processed using proprietary software against a local base receiver whose coordinates are taken as known. The relative position vector is quite good if the distance between is less than a few km.

OPUS and some of the proprietary processing methods are known as Static, because you leave the receiver in place for a while.  Some of the processing methods are known as RTK (real-time kinetic) because a radio link brings the fixed-station data to the receiver for immediate processing.

There is no way you will ever get good enough results with amateur equipment to get a mark into the NGS data base.  Their procedure is called "Bluebooking" after the original manual for the process.  It requires extensive measurements (multiple 4-hour sessions?) with professional grade L1/L2 equipment and a lot of arcane file submissions. They expect to be repeatable within a cm or two.

Let's be very clear, "NGS support for the database of survey marks" is NOT going away.  In fact the agency is actively working on new provisions where surveyors (and others) who want to follow the data submission protocols - called Blue Booking and/or use OPUS Projects and Shared Solutions can easily submit their own data for publication.  NGS takes this effort very seriously.  What has changed is that these passive marks will no longer define the horizontal and vertical datums of the National Spatial Reference System, that will be performed by the network of Continuously Operating Reference Stations (CORS).  Passive marks will continue to play a significant role in land surveying and many engineering and geophysical applications.

Information on how to properly set a passive mark can be found in NGS BENCH MARK RESET PROCEDURES beginning on page 14.  Bill93 is absolutely correct, you will never be able to meet the accuracy requirements for NGS with a simple cheap code receiver.

If the CORS network is going to define the datum, then doesn't that mean that the satellite positions are ultimately what defines the datum? How do we know where the satellites are?  If the whole purpose of the NSRS is to define and determine positions on the surface of the earth it makes more sense to me to have everything ultimately depend on things that are actually stuck in the earth. I feel unmoored, and all at sea ...

You're getting into satellite positioning and other stuff - those satellites know exactly where they are, at all times.

You want things stuck on the earth - but, the tectonic plates are shifting, moving. So even if everything was 'fixed' on earth here, they woldn't actually be 'fixed'.

BTW, Dave - I thought you were retired! What the hell you doin', man?

As far as static markers still being used - that's a huge positive there. I just had to flag for a survey crew along our tracks - 20 miles of ROW being surveyed for elevation and ditch grading. Readings every 30 feet. Every morning they would set up on a nearby NGS level mark, run to the railroad and back, verify the measurements, then take off heading down the tracks. I learned a lot getting to watch them do, what I call, the "other side" of benchmarking: Finding a station, using it for a baseline elevation, then reading offset readings every x feet.

I did an experiment with a couple of (relatively) inexpensive RTK receivers (uBlox F9P) and the RTKLIB software, to locate a point in my back yard to a precision of about 2 cm.  Fortunately, there is a NGS horizontal control about 1.5 km from my home, and I put one receiver over my mark in the back yard, and the other over the NGS mark and let it gather data for about a half hour.  Then I put the data in to the RTKLIB post-processing software and got a position for my mark.  I repeated the process several times over several weeks, and kept getting the same results within about 1 cm.  My goal was to be able to set up a base station in my back yard that doesn't risk getting stolen, and use the rover around my neighborhood for accurate mapping data.

Each receiver cost about \$250 dollars, and I built them myself from parts.  It was largely a project to entertain myself.  I think you can buy a turnkey receiver for about \$400-\$500 dollars.

2 hours ago, holograph said:

Each receiver cost about \$250 dollars, and I built them myself from parts

Wait, did the receivers cost \$250, or the parts FOR the receivers cost \$250?

The parts cost that.  I also used some parts on hand, so I guess if you had to buy everything it would be maybe \$325 in parts, each, excluding the antennas.  You can buy similar assembled and ready-to-go receivers for \$400-\$500 (each) from Sparkfun, without antennas.

The uBlox F9P module is a dual frequency L1/L2 receiver that can be configured as either an RTK base or rover.  As a base, it can generate RTCM 3 messages, and as a rover it can read those messages and compute a RTK solution.  I wrote my own communications software interface using an ESP32 module (about \$8) that doubled as a memory card data logger, captured the raw pseudo-range data, and then I post-processed it.  My UI is still a rough prototype and I haven't been motivated enough to put together a user-friendly RTK system.

Basically, my base station is battery powered and connects to my home WiFi network.  My rover is also battery powered and connects to my cell phone using Bluetooth.  The batteries are off-the shelf 5v phone-chargers that can power the receivers for 24 hours or more.  I've attached images of the base station unit and the rover unit.  They're not as compact as commercial versions, but they were fun to build.

Nice to see you posting - I hadn't noticed you on here for a looooong time.  Also good to see someone working with the technical side.

On 8/31/2021 at 3:50 PM, holograph said:

The parts cost that.  I also used some parts on hand, so I guess if you had to buy everything it would be maybe \$325 in parts, each, excluding the antennas.  You can buy similar assembled and ready-to-go receivers for \$400-\$500 (each) from Sparkfun, without antennas.

The uBlox F9P module is a dual frequency L1/L2 receiver that can be configured as either an RTK base or rover.  As a base, it can generate RTCM 3 messages, and as a rover it can read those messages and compute a RTK solution.  I wrote my own communications software interface using an ESP32 module (about \$8) that doubled as a memory card data logger, captured the raw pseudo-range data, and then I post-processed it.  My UI is still a rough prototype and I haven't been motivated enough to put together a user-friendly RTK system.

Basically, my base station is battery powered and connects to my home WiFi network.  My rover is also battery powered and connects to my cell phone using Bluetooth.  The batteries are off-the shelf 5v phone-chargers that can power the receivers for 24 hours or more.  I've attached images of the base station unit and the rover unit.  They're not as compact as commercial versions, but they were fun to build.

Ah, awesome. Some newer Android phone chipsets are starting to allow dual-band readings as well, though I am not sure about their accuracy. I am looking into it for some additional features to my Android app, but honestly? I'm a hardware guy, and love having the hardware at my disposal.

## Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

×   Pasted as rich text.   Paste as plain text instead

Only 75 emoji are allowed.