Allan
Well-Known Member
I'm equally as impressed as the OP. It's amusing to think that some of the surveys are hundreds of years old!
Allan
Allan
The accuracy of GPS/Navionics just blows my mind sometimes...
- Thread starter Andrew_Trayfoot
- Start date
GHA
Well-Known Member
Another another
, just incredible that a little chip costing a few dollars can recieve low power radio signals from far above, decode into data and calculate with such accuracy from the tiny differences in time for the signal to arrive from tbe satilitte , plus factored is the mass of the earth and the speed of the satellites changing just how fast time itself is for the satellites compared to us. Jaw dropping really.Back onboard, sasplanet is great to check chart datums, jump between navionics/cmap and various satilitte image sourves with keyboard shortcuts to compare.
AntarcticPilot
Well-Known Member
No, it would have been a much older datum relating to a local survey. At the scales we're looking at WGS72 is imperceptibly different from WGS84.
ghostlymoron
Well-Known Member
Yes you're right, pilot. It's WGS84. The original survey was no doubt related to an astro nav position taken by someone wearing a three cornered hat!
AntarcticPilot
Well-Known Member
Well, the "cocked hat" might have been quite small! ����
ghostlymoron
Well-Known Member
ghostlymoron
Well-Known Member
There's no comparison between GPS and traditional methods dead reckoning is pretty inaccurate, astro only tells you where you were half an hour ago. Astro is a good way of filling time on a passage though.
Andrew_Trayfoot
Well-Known Member
Going back to my track.
The channel in Lilliput Lagoon dries twice a day, and only has 1.75m of water in it at the top of a very high spring. No commercial traffic will ever go in there and even most LSC cruisers use it twice a year. Yet it is bang on. You can even see where I went the wrong side of a post in the dark.
I might add the posts to the chart one day, if I've got nothing better to do. Our POSH post is already there on the Community Laye.
As I said amazing...
The channel in Lilliput Lagoon dries twice a day, and only has 1.75m of water in it at the top of a very high spring. No commercial traffic will ever go in there and even most LSC cruisers use it twice a year. Yet it is bang on. You can even see where I went the wrong side of a post in the dark.
I might add the posts to the chart one day, if I've got nothing better to do. Our POSH post is already there on the Community Laye.
As I said amazing...
Buck Turgidson
Well-Known Member
In these discussions it's important to distinguish between the accuracy of the fix and the accuracy of the chart you are overlaying it on.
My garmin explorer+ was on for 19days continuously in July from Falmouth to Valencia and the largest lateral inaccuracy it calculated was +/- 4m. It was using SBAS throughout. Never lost either navstar or EGNOS signal once.
In my day job however we have problems approaching Cairo airport every day as the Russians are jamming GPS in the region.
My garmin explorer+ was on for 19days continuously in July from Falmouth to Valencia and the largest lateral inaccuracy it calculated was +/- 4m. It was using SBAS throughout. Never lost either navstar or EGNOS signal once.
In my day job however we have problems approaching Cairo airport every day as the Russians are jamming GPS in the region.
jwilson
Well-Known Member
I think it's the charts that are out, not the GPS. I've watched a plotter track show us sailing across the middle of a substantial island in the China Sea as we sailed past it.
Now SA is turned off the positions from the two fixed GPS receivers on my boat regularly differ by .001 mile - the six feet between the two mushrooms. Once I've been somewhere in good visibility and seen that the charts match the GPS I think you can be confident the charting is good - never met anywhere in the UK that's way out.
AntarcticPilot
Well-Known Member
A discourse on cartographic accuracy!
It is a truism that all maps and charts are inaccurate in some way - this is unavoidable, because maps and charts are abtractions of reality, not reality itself - the well known phrase "the map is not the territory" sums this up very concisely!
Charts and maps have inaccuracies that derive from many different sources, but in general they fall under one of the following headings:
1) Absolute errors in position fixing. A map may be internally consistent in the relationship between features, but be misplaced in absolute terms. Before the advent of Very Long Baseline Intereferometry(VLBI), pretty much all national surveys were independent and related back to astronomical observations which varied from single star-fixes taken by surveyors using theodolites (best accuracy in absolute terms a few hundrend metres) to networks of astronomical observations including fixed, long-term observatories (e.g. the OS National Grid) where the position fixing was good to perhaps tens of metres. In remote parts of the world, maps are still related to astronomical fixes, and so we should expect errors in the region of hundreds of metres with respect to GPS even if the survey is internally consistent.
2) Internal errors in survey reduction. Error is a bit strong, but that's the end result! Prior to the widespread use of accurate satellite position fixing, Surveyors would fix the position of locations such as monuments ("trig points") by measuring angles between them. The resulting network was then computed to minimize the residual error at each surveyed point. In most cases this meant that if you measured the position of a point from one survey point, it would agree with the location computed from a different survey point. However, altough this was internally consistent, it deviated from global datums like WGS84 by an amount that varied irregularly from place to place and (because OSGB36 was computed in two parts) could have sudden changes in the direction of the contours of equal difference. The deviations from a global datum like WGS84 are of the order of 100m; usually less. We aren't likely to come across this class of errors in modern maps of developed countries, but in places with less investment in infrastructure. it will still be a common source of error, especially in archipelagoes, where it is likely that the internal survey links within an island are much stronger than the links betwen islands (cartographers like to "lose" survey errors in places that don't have much on them - like channels between islands!)
3) Errors in interpretation of the terrain. Again, not a problem in inhabited countries, but very much a problem in places where there a) isn't a population, or at least not one that contributes to maps, and b) where the terrain is either unusual or changeable. Both are true of Antarctica, where I did most of my mapping! But imagine the problems of translating mapping practice adapted to the UK to a country with mangrove swamps, where the position of the "shoreline" mat be uncertain by large amounts, or Antarctica where there are whole classes of features that simply don't exist anywhere else (I had a long exchange with one project to create a global map where they had trouble understanding that ice-shelves don't correspond to any feature that exists in temperate countries, and which make the definition of "land" and "sea" a bit tricky (an iceshelf is floating ice derived from land-based glaciers, where the ice may be from hundreds to thousands of metres thick - to the traveller they are land, but they float on sea water)! And even where mapping conventions have been developed to represent the relevant features, the map or chart may well be compiled by someone without experience of the terrain they are mapping; I've come across quite a few places in Antarctica where it is clear that the original mapper didn't understand the nature of what they were mapping.
4) Errors of observation. Rare in these days of satellite images, but common in earlier mapping. My favourite examples are from Cook's mapping of New Zealand - he mapped "Stewart Peninsula" and "Banks Island" - of course, we now know that it is "Stewart Island" and "Banks Peninsula". Even the great man made mistakes - how many more have other, lesser map-makers made over the centuries?
5) Errors of transcription. This is the commonest form of error in electronic charts other than the original data from Hydrographic Organizations. People like Navionics etc. do not carry out their own survey; they buy data from the appropriate Hydrographic Organizations and reformat it for their customer base, adding value as they see fit. Sadly, it does appear that this reformatting sometimes introduces error, either because mistakes have been made in the data conversion (which can result in unclosed "loops" in coastlines or contours), or mistaken attribution meaning that a feature is in the right place but has the wrong attrributes (things like spurious drying areas have been noted here).
6) Generalization. This isn't an error, but it is important to be aware of it. As I mentioned at the beginning, all maps are abstractions of reality, and the representation of reality is chosen to be appropriate for the scale of the map. So, a large-scale map might show indivdual rocks, but a small-scale map will show a reef, and so on. This isn't an error, but it is a change in the way reality is being represented, and all map users should be aware of the scale of the data they are using. Things like contours and coastlines will be smoother and more general on a smnall scale map than on a large scale one, lesser marks may be omitted on a small-scale map, and so on.
7) Crowd-sourcing is particularly subject to error, because it is very difficult to apply QA to crowd-sourced data in areas where data are sparse. It is useful in areas where there are many contributors, or where contributions are received from well-equipped and conscientous observers, but the quality of crowd-sourced data may be extremely variable. Many examples of the pitfalls of crowd-sourced data have been presented in thise fora; notably in cases where data are sparse.
8) Derived features. Every user of charts should be aware that some features commonly presented on charts are NOT primary data, but derived. Contours are the most widely seen examples; the data are the depth soundings, NOT the contours, which are derived from the depth soundings. In well-surveyed areas the contours may be based on more data than are shown on the face of the chart, and hence quite reliable, but in other areas they may well be a "join the dots" exercise that misses sudden changes of depth. This is a particular problem with crowd sourced data where the depth contours will have been generated entirely automatically and where the trracks of contributing boats may well be constrained in ways that give a bad distribution of data.
9) Human error - someone made a mistake! It happens, and someone posted a nice example on these fora where Garmin had published a map where the code for a rocky bottom had been interpreted as being isolated rocks!
The take home message is NEVER rely on the chart alone; the best guide is the chart coupled with your own observations. If you can see breaking water, then it is unlikely that the water is deep, even if the chart says it is!
It is a truism that all maps and charts are inaccurate in some way - this is unavoidable, because maps and charts are abtractions of reality, not reality itself - the well known phrase "the map is not the territory" sums this up very concisely!
Charts and maps have inaccuracies that derive from many different sources, but in general they fall under one of the following headings:
1) Absolute errors in position fixing. A map may be internally consistent in the relationship between features, but be misplaced in absolute terms. Before the advent of Very Long Baseline Intereferometry(VLBI), pretty much all national surveys were independent and related back to astronomical observations which varied from single star-fixes taken by surveyors using theodolites (best accuracy in absolute terms a few hundrend metres) to networks of astronomical observations including fixed, long-term observatories (e.g. the OS National Grid) where the position fixing was good to perhaps tens of metres. In remote parts of the world, maps are still related to astronomical fixes, and so we should expect errors in the region of hundreds of metres with respect to GPS even if the survey is internally consistent.
2) Internal errors in survey reduction. Error is a bit strong, but that's the end result! Prior to the widespread use of accurate satellite position fixing, Surveyors would fix the position of locations such as monuments ("trig points") by measuring angles between them. The resulting network was then computed to minimize the residual error at each surveyed point. In most cases this meant that if you measured the position of a point from one survey point, it would agree with the location computed from a different survey point. However, altough this was internally consistent, it deviated from global datums like WGS84 by an amount that varied irregularly from place to place and (because OSGB36 was computed in two parts) could have sudden changes in the direction of the contours of equal difference. The deviations from a global datum like WGS84 are of the order of 100m; usually less. We aren't likely to come across this class of errors in modern maps of developed countries, but in places with less investment in infrastructure. it will still be a common source of error, especially in archipelagoes, where it is likely that the internal survey links within an island are much stronger than the links betwen islands (cartographers like to "lose" survey errors in places that don't have much on them - like channels between islands!)
3) Errors in interpretation of the terrain. Again, not a problem in inhabited countries, but very much a problem in places where there a) isn't a population, or at least not one that contributes to maps, and b) where the terrain is either unusual or changeable. Both are true of Antarctica, where I did most of my mapping! But imagine the problems of translating mapping practice adapted to the UK to a country with mangrove swamps, where the position of the "shoreline" mat be uncertain by large amounts, or Antarctica where there are whole classes of features that simply don't exist anywhere else (I had a long exchange with one project to create a global map where they had trouble understanding that ice-shelves don't correspond to any feature that exists in temperate countries, and which make the definition of "land" and "sea" a bit tricky (an iceshelf is floating ice derived from land-based glaciers, where the ice may be from hundreds to thousands of metres thick - to the traveller they are land, but they float on sea water)! And even where mapping conventions have been developed to represent the relevant features, the map or chart may well be compiled by someone without experience of the terrain they are mapping; I've come across quite a few places in Antarctica where it is clear that the original mapper didn't understand the nature of what they were mapping.
4) Errors of observation. Rare in these days of satellite images, but common in earlier mapping. My favourite examples are from Cook's mapping of New Zealand - he mapped "Stewart Peninsula" and "Banks Island" - of course, we now know that it is "Stewart Island" and "Banks Peninsula". Even the great man made mistakes - how many more have other, lesser map-makers made over the centuries?
5) Errors of transcription. This is the commonest form of error in electronic charts other than the original data from Hydrographic Organizations. People like Navionics etc. do not carry out their own survey; they buy data from the appropriate Hydrographic Organizations and reformat it for their customer base, adding value as they see fit. Sadly, it does appear that this reformatting sometimes introduces error, either because mistakes have been made in the data conversion (which can result in unclosed "loops" in coastlines or contours), or mistaken attribution meaning that a feature is in the right place but has the wrong attrributes (things like spurious drying areas have been noted here).
6) Generalization. This isn't an error, but it is important to be aware of it. As I mentioned at the beginning, all maps are abstractions of reality, and the representation of reality is chosen to be appropriate for the scale of the map. So, a large-scale map might show indivdual rocks, but a small-scale map will show a reef, and so on. This isn't an error, but it is a change in the way reality is being represented, and all map users should be aware of the scale of the data they are using. Things like contours and coastlines will be smoother and more general on a smnall scale map than on a large scale one, lesser marks may be omitted on a small-scale map, and so on.
7) Crowd-sourcing is particularly subject to error, because it is very difficult to apply QA to crowd-sourced data in areas where data are sparse. It is useful in areas where there are many contributors, or where contributions are received from well-equipped and conscientous observers, but the quality of crowd-sourced data may be extremely variable. Many examples of the pitfalls of crowd-sourced data have been presented in thise fora; notably in cases where data are sparse.
8) Derived features. Every user of charts should be aware that some features commonly presented on charts are NOT primary data, but derived. Contours are the most widely seen examples; the data are the depth soundings, NOT the contours, which are derived from the depth soundings. In well-surveyed areas the contours may be based on more data than are shown on the face of the chart, and hence quite reliable, but in other areas they may well be a "join the dots" exercise that misses sudden changes of depth. This is a particular problem with crowd sourced data where the depth contours will have been generated entirely automatically and where the trracks of contributing boats may well be constrained in ways that give a bad distribution of data.
9) Human error - someone made a mistake! It happens, and someone posted a nice example on these fora where Garmin had published a map where the code for a rocky bottom had been interpreted as being isolated rocks!
The take home message is NEVER rely on the chart alone; the best guide is the chart coupled with your own observations. If you can see breaking water, then it is unlikely that the water is deep, even if the chart says it is!
Ric
Well-Known Member
Re: A discourse on cartographic accuracy!
I work in mapping too, flying a survey aeroplane fitted with LIDAR and multiple high resolution cameras. Most of our projects are in Africa, and it amazes me how accurate the original Victorian surveys were, using just theodolites, compasses, dip-circles and sextants. The maps they created are still widely used today.
We still battle to make decent maps in Africa. Most of our GCPs (Ground Control Points) get destroyed by the locals within a few days because they think somebody is trying to steal their land. The resilience of the original surveyors in Africa was astounding...
I work in mapping too, flying a survey aeroplane fitted with LIDAR and multiple high resolution cameras. Most of our projects are in Africa, and it amazes me how accurate the original Victorian surveys were, using just theodolites, compasses, dip-circles and sextants. The maps they created are still widely used today.
We still battle to make decent maps in Africa. Most of our GCPs (Ground Control Points) get destroyed by the locals within a few days because they think somebody is trying to steal their land. The resilience of the original surveyors in Africa was astounding...
Last edited:
lw395
Well-Known Member
Re: A discourse on cartographic accuracy!
The quantity of work they must have done is astounding. Must be big numbers of survey points, lots of calcs and checks?
All processed longhand?
Plus some real skill to visualise what they were drawing.
The quantity of work they must have done is astounding. Must be big numbers of survey points, lots of calcs and checks?
All processed longhand?
Plus some real skill to visualise what they were drawing.
AntarcticPilot
Well-Known Member
Re: A discourse on cartographic accuracy!
The problem (pre-GPS) in Antarctica was that survey was an extra done on an opportunistic basis. A party went into the field to (say) study the geology of a new area and was expected to carry out survey work as the opportunity arose. Although every field party included people who were trained in survey techniques, it was rare to have a dedicated surveyor - they were mostly scientists who had done a crash course in Cambridge. Some took it seriously and did an excellent job; others didn't. And the survey network wasn't designed in advance to provide the best cover and connectivity. The result was that when a major recalculation of the survey network took place in 1988, a vast number of observations had to be rejected as being of too poor a quality or not providing any extra value in the network. And in places the network was very weak; long chains of observations without adequate triangulation. Even so, the RMS error in the recalculated network, which used Transit satellite observations as the datum, was only 11 metres.
In former UK colonies, the survey would have been carried out by the Department of Overseas Surveys, which was absorbed into the OS in the 1990s. It would have been a fully professional job, so it isn't surprising the results were first class. The DOS was very reluctant to get involved in Antarctic mapping because they knew the survey was not to the standards they expected elsewhere.
The problem (pre-GPS) in Antarctica was that survey was an extra done on an opportunistic basis. A party went into the field to (say) study the geology of a new area and was expected to carry out survey work as the opportunity arose. Although every field party included people who were trained in survey techniques, it was rare to have a dedicated surveyor - they were mostly scientists who had done a crash course in Cambridge. Some took it seriously and did an excellent job; others didn't. And the survey network wasn't designed in advance to provide the best cover and connectivity. The result was that when a major recalculation of the survey network took place in 1988, a vast number of observations had to be rejected as being of too poor a quality or not providing any extra value in the network. And in places the network was very weak; long chains of observations without adequate triangulation. Even so, the RMS error in the recalculated network, which used Transit satellite observations as the datum, was only 11 metres.
In former UK colonies, the survey would have been carried out by the Department of Overseas Surveys, which was absorbed into the OS in the 1990s. It would have been a fully professional job, so it isn't surprising the results were first class. The DOS was very reluctant to get involved in Antarctic mapping because they knew the survey was not to the standards they expected elsewhere.
Last edited:
tudorsailor
Well-Known Member
If you want a precise way of relaying your location you should have a look at https://what3words.com/about-us/
TudorSailor
AntarcticPilot
Well-Known Member
What's wrong with latitude and longitude that you can read directly from your chart or GPS, and which is potentially as as accurate as whatever means of poisition fixing you're using?Whatthree words is of use on land where geographic coordinates are rarely available or used, but at sea I can't see the point.