GPS drifting ? At anchor

[StuM]

I agree with those who say that plot looks like anchor drag with changing tide, not GPS error.

[Olddave]

Originally Posted by Dave_S

I spent most of last night trying to work out if my anchor was dragging.

I was anchored in 2-4 meters of water on clean sand in strong current with 15 knots of breeze with 20m of chain. I set the anchor alarm location by scope and direction after I set the anchor rather than when i dropped it.

The alarm went off and the trace looked like tide change, I checked my visual markers and it looked Ok. I thought I had screwed up my anchor location so I reset it to the middle of the trace.

The alarm kept going off and the trace kept getting bigger and bigger, I don't like to blame the equipment when things don't match up but it said I had drifted over 180m and my visual markers didn't look like I had moved much but it was a moonless night. So after a dodgy night's sleep I awoke to find I hadn't dragged but that the equipment was wrong.

Does GPS do this sometimes or is the tablet I'm using crap at reading the satellites or is "anchor light" software crap?

Dave S

As you first thought. You were dragging, and you can see the tell tale signs in the track recorded.
Classic dragging signature. Slide backwards and then stabilise for a while, then off again.
I recommend using more chain as a standard approach to anchoring. Or rope if that’s what you have, but chain is the ants pants.

In 4m of water, I would put out at least 40m of chain and more if there was concern about current and wind combining. Be careful in places where strong tides may be washing the bottom so that you only have a thin layer of sand over a rock bottom. You can look at the chart and predict this quite often. Very obvious in a gutter between islands or where the water bends around a headland or point etc.
If you haven’t done any formal Nav studies it is a good idea because you will pick up tips that last a lifetime.
Dave

[Mike Banks]

Clean sand is just about one of the worst holding ground there is. Only loose gravel could be worse. In this ground you need to employ a large fluked anchor and use a kellet with plenty of chain out. I had the same thing in clean sand. It is just a matter of using a big anchor and a low angle of strain on it.

[noelex 77]

Originally Posted by kish

Are you aware that the inherent accuracy of the gps signal is 15 metres? Using differential gps the accuracy can be improved, but this is not the case on gps receivers at consumer prices.

This is the type of accuracy you should expect over 24 hours.

This is an actual GPS plot from a marine unit over 24 hours. This is GPS alone without GLONASS or a satellite differential fix (although GLONASS is a less accurate system that helps redundancy, but not accuracy) .

You can see there are occasional excursions out to 4m even 5m over longer time periods, but 15m is too pessimistic.

Another way of looking at the accuracy is this plot shows the position is accurate to better than 2m 94% of the time. When looking at the GPS plot on the chartplotter the position can be averaged over a longer time period so detecting dragging of 1-2m is possible with a modern GPS that is mounted where it has a clear view of the sky especially with a differential fix.

Unfortunately, on many boats the GPS unit is not well located and the errors can be much greater. A well installed GPS can be as accurate as transits, perhaps more so, with the advantage that it still works in poor visibility and situations where the boat is moving around the swinging circle.

[sazarac]

Originally Posted by sazarac

Was your altitude all over the place also?

Since you are supposed to be at sea level and your gps measures x,y and z, if one plane of reference seems wrong, x and y because of suspicions regarding GPS accuracy the Z value should be fluctuating as well as the x and y. Nice way to get an error range for x and y.

[Wotname]

Originally Posted by sazarac

Since you are supposed to be at sea level and your gps measures x,y and z, if one plane of reference seems wrong, x and y because of suspicions regarding GPS accuracy the Z value should be fluctuating as well as the x and y. Nice way to get an error range for x and y.

Please remember that the accuracy of the x & y values are much greater than the z value. Mainly due IIRC, to the geometry of the satellites.

[stocru]

To check if dragging put hand on rode. Usually you can feel the movement and often get an idea of the type of bottom. For dragging alarm lower the weight on your leadline from the after end of the boat (the ill equipped can use anything heavy - not a container of water) to the bottom, give it some slack to allow for movement, tie the inboard end to several saucepans placed on a cockpit side bench. Drag and the crash should alert you.

[mikedefieslife]

GPS does drift. Just stand on land and turn on a Garmin device with a 10hz polling rate. You'll see it move around. Add to that the built in in-accruacy of GPS courtesy of the US military and you'll find things aren't as precise as they could be. Turning on GLONASS if your device supports it in addition to GPS could help.

Will be better once we have Galileo up and running properly, which will be accurate to within a metre for personal use

[Dave_S]

I took a screen shot of the trace from my desk at work today _ about 5 hr's. Doesn't look the same but doesn't look good. Could be the problem. The scale at the bottom right is 100'. So about 700' of movement (don't count the line leaving the page, that's when I drove off.

We are about 27° 42 686s it said it can see 11 satellites but only used 3-4.

I have a rockna - vulcan 33kgs and all chain to a nylon bridle.

I used the red and green markers - the green was about 250m North and the red about 300m 40° to it's west. There was a boat about 300m 10° east of the green. At no time did the 40° become more than 45° even when we were supposed to be at the closest point.

It is quite possible I screwed up the anchor position, I reset it after looking at the early trace that looked like two rest positions with the anchor in the middle but this failed.

The + is a mark for people to add notes, the anchor would be somewhere under the first couple of letters in the word anchorage.

[noelex 77]

Originally Posted by mikedefieslife

Turning on GLONASS if your device supports it in addition to GPS could help.

I think having a receiver that can recieve GLONASS is sensible, but it does not seem to add to the accuracy. For example, in the test I reported in post #34, the accuracy actually dropped when GLOSASS was enabled in addition to the standard GPS (the better than 2m accuracy dropped from 94% to 79%).

No doubt better algorithms could have integrated the GPS and GLONASS position more effectively, but at least currently GLONASS does not seem to offer much promise for better position accuracy if a good GPS signal is available. At least that is my understanding.

[Cavalier]

Originally Posted by Dave_S

I took a screen shot of the trace from my desk at work today _ about 5 hr's. Doesn't look the same but doesn't look good. Could be the problem. The scale at the bottom right is 100'. So about 700' of movement (don't count the line leaving the page, that's when I drove off.

We are about 27° 42 686s it said it can see 11 satellites but only used 3-4.

I have a rockna - vulcan 33kgs and all chain to a nylon bridle.

I used the red and green markers - the green was about 250m North and the red about 300m 40° to it's west. There was a boat about 300m 10° east of the green. At no time did the 40° become more than 45° even when we were supposed to be at the closest point.

It is quite possible I screwed up the anchor position, I reset it after looking at the early trace that looked like two rest positions with the anchor in the middle but this failed.

The + is a mark for people to add notes, the anchor would be somewhere under the first couple of letters in the word anchorage.

This is some good work! Often, folk post question like yours on this forum and then sit back and wait for an answer, which is fair-enough. Not so often do people continue their own detective work, trying to debug or make sense of their data, and then report back to the forum with their reasoned learning.. bravo, sir!

Looking at the data of your latest post, along with the GPS log of your initial post, and considering the most likely current conditions etc at that physical location, my best guess would be that your initial GPS log was a result of a mixture of some dragging combined with the oddities of your specific GPS unit.

As some have pointed out, one would expect a star-burst pattern is the GPS was located at a single point and started to 'wonder'

However, with a boat on the hook, and with a decent amount of scope, you'd expect this to 'blur' a bit and turn into the same star-burst distributed around a ring.

Where you were located on that section of coast, your swing around the hook would be been much more elliptical than circular (at change of tide) - add in some drag in both directions up and down the coast, and the result would be very similar to what you can see in your initial GPS log.

My earlier comment regarding the 'mission' profile of a GPS unit is based more upon the software than the actual GPS unit. Many of the gps receivers that are found in consumer electronics are basically the same mass-produce commoditized package (that comes with an SDK) used by many hardware producers.. ultimately, they're packaging together available electronics into a new package, rather than developing every single processor/ASIC etc from scratch. It's the software implementation that makes all the difference. For example, a GPS software implementation used specifically for road use will always assume that your altitude/elevation must be very near or upon the known topographical map/data for the specific geophysical location you're at. You'll also notice that the software attempts to keep you on the closest road.. this can be seen clearly when you're driving on a feeder, for example, and software still thinks you're on the main freeway. You then take off in a different direction on a curve (or the freeway curves away) and for a couple of seconds the software insists that you're still on the freeway, until your physical position comes outside of the probable error circle that was set in software - at which point it'll 'pin' you to the closest possible solution which is the road you're actually on.

Each software implementation of a GPS logging/tracking product will make various assumptions based upon the developer's assumptions about its use.

For an anchor watch software, my personal assumption would be that 'altitude' is always near zero, so "z" would be disregarded (unless there was a benefit to looking at that w.r.t. error/QA of the signal).. and i'd probably have quite a long period set for my rolling average, to help filter out the spurious errors in position and stop logged points falling outside of my set 'alarm circle'.

However, the person who developed this software may have been sat in their house at the top of a hill within 30deg of the equator, for all you know, and their testing showed considerably less wondering by virtue of being able to see more of the sky, and more of the satellites!

On that note, if you happen to find yourself at anchor in a similar location as before, and the weather is good enough, it may be worth putting the tablet/device that you were using up on the coach-roof and see whether a bit of elevation with an overall better signal leads to less wondering over-night.

[ZULU40]

Originally Posted by Dave_S

Does GPS do this sometimes or is the tablet I'm using crap at reading the satellites or is "anchor light" software crap?

Of course I dont know what device you are using, but I do know something about ipad. There are two kinds of ipad, the 3 and 4G models, and the simply wireless models.

Interestingly wireless models are still fitted with a GPS chip so can receive GPS data, but without the resolve to poll fone towers they lack a great deal of accuracy.

A 3G or 4G doesnt need to be connected to a subscribed service to have 10x the GPS acuity of purely wireless models, they will still poll towers. The difference normally amounts to an accuracy of 20-30ft or so for 3G and 4G, but more like 300ft+ for purely wireless devices.

All the above applies to those tempted to use an ipad for charting purposes too.

Hope that helps

[SURV69]

I have never seen the kinds of errors I see posted .... at least not after the initialization.

I have worked with nothing but survey-grade GPS and precision instruments since 1998. EVERYTHING I've done since then is in Ohio State Plane system Nad83, or newer, with precision generally in the 1/4" arena ... often much better.

I always noticed that there were some sats that were slightly north of the 42 parallel, no doubt to give Alaska suitable coverage.

Now, I have no idea what kind of accuracy can be expected on consumer-grade GPS units, in the southern hemisphere, but being south would have to mean that all the sats transmissions would have to go through a lot more atmosphere.

I think I know(?), that consumer-grade GPS never or almost never, allows masking. Here, in the northern hemisphere, we always mask around 40 degrees, sometimes less and often more. This ensures that those more questionable sats will not be used for computation by the instrument.

Up here, we generally use GLONAS for quicker initialization, or if we find ourselves in trees ... maybe.

Are there sats systems that better service the southern hemisphere, or are many boaters stuck with the more northernly U.S. GPS system?

BTW: in the old days I used to employ actual elevation compared to individual GPS readings as a sort of benchmark to help me determine what readings were most likely to be the most accurate ... elevations, even though they too are very(relatively speaking), accurate, elevation readings always suffer.

[a64pilot]

Originally Posted by Wotname

Please remember that the accuracy of the x & y values are much greater than the z value. Mainly due IIRC, to the geometry of the satellites.

I’ve tried several times to use the altitude feature of GPS with what is known as highway in the sky, Altitude is unreliable, I believe in part due to that the GPS software assumes a round Earth, and the Earth isn’t.
Best guaranteed accuracy for GPS back in the day when selective availability was enabled was 100 meters, Bill Clinton had selective availability turned off in order to make the US GPS system more accurate than GLONASS, which it isn’t with SA on. Current DGPS is stated to be within 10 meters.
A concern was with a GPS anyone could make a poor man’s guided Missile, and in fact if you export an aircraft with a GPS you may have to prove that it won’t work above x speed, I don’t know that number but did have to prove to the CBP that an aircraft I was exporting with dual Garmin 430’s couldn’t be used as a missile guidance system, something that although I have never thought about it until then, had assumed it could, but Garmin was well prepared to supply that proof, they had been there before obviously.
SA can be turned on at any time there is the need, assumption is to deny an enemy accurate GPS, Military GPS uses the “P” code of the satellites which requires a secure fill and it’s accuracy is incredible compared to civilian GPS which uses the “C” code. Stronger signal to penetrate jungle foliage etc too.
It was Ronald Reagan who after Korean Airline Flight 007 who decided to make GPS freely available to the public.
The US government has and I assume will continue to deny GPS coverage to certain areas at certain times, Indian war in 1999 I know was admitted to.
To make GPS very accurate like to position farm tractors or have an aircraft shoot an approach a ground based Station was needed, then later when DGPS System was launched, the need of a ground based system was reduced.

Long ago when people first started flying with IPads with charts on them it was discovered that the built in Apple GPS isn’t very good, for flying you need an external GPS, maybe not so much on a 6kt Boat, your decision.

My first handheld GPS back maybe over 20yrs ago was an Apollo Precedus, it was interesting in that it had a survey mode, idea was that you place it in one spot and it would continue to take measurements and over time, it’s position got more and more accurate as SA and inaccuracies would be averaged out. It was terribly accurate when compared to known surveyed positions if allowed to take and average hundreds of positions.
I’ve often wondered why no other GPS I have seen has that feature?

[hellosailor]

"To set the record straight a GPS receiver requires at least four satellites to come up with a position solution."
True but deceptively misleading. As kas mentioned, you only need 3 satellites for a 2 dimensional fix, and "sea level" being somewhat all that sailors need, that's all our GPSes need. Unless you are using the GPS on a submarine or other vessel that might not be on the surface.

What no one has mentioned is that GPS signals, which arrive by radio waves trying to bash their way through a constantly changing atmosphere, "twinkle" the same way that light coming form the stars twinkles. The constant interference means there is always a shift in the signals due to local interference, air currents from heat where the oceans meet the land being some of the most common and unpredictable.
That's why you will see mention of WAAS and EGNOS systems, where a separate signal is used to determine the probable local atmosopheric distortions to the GPS signals, and that compensation figure is also broadcast to receivers so they can apply it to their calculations.
If there is no WAAS or EGNOS equipment in your area (in the US it was mainly one on the east coast, one on the west coast, nothing in Kansas or the middle) or if your receiver (tablet) lacks the ability to USE that extra signal, you will find your position error drifts way more than a proper GPS unit with that ability built in.
If there are any weather fronts coming through, with corresponding changes in the air temperatures, or there is daylight/sunset causing thermals, again, you'll see more errors.

"Survey mode" sounds deceptive too. It sounds like position averaging, which has been done. But it is not real "survey" anything. Surveyors use differential GPS with extra enhancements to get sub-centimeter accuracy--at a price.