Inaccurate RYA Teaching : CTS - Quest For a New Method

[Dockhead]

Originally Posted by goboatingnow

It is not DockHead. Its gives a more correct result on paper, and only so if one applies the axiom "The tide remains the same for the hour under consideration".

IN reality it doesnt and the values interpolated from the tide tables often have significant lack of precision in them. Thats makes SWL method no worse or better then any other

It assumes the tides behaves AT THE AVERAGE during one hour. But inflating the vector triangle applies the average of THE WHOLE PASSAGE. Order of magnitude (roughly) more error. I can't believe you don't see this. C'mon Dave, don't be vain.. We've all made mistakes and said stupid things on this thread, from which we've all learned so much

[goboatingnow]

Its a fools gold to teach students increasing precise calculations on original imprecise data.

If you want correct precision, that replot ALL the tides with the time interval decreasing towards the greatest precision the tide data is available for. Thats precision.

Not what were talking about.

[LJH]

Originally Posted by goboatingnow

Ive said this several times on the other thread,

The RYA method only assumes at the very worst a +- 30 minutes "inflation or deflation" Secondly all good instructors point out the case where you are right on or near the 30 minute boundary. The RYA method does not blindly "inflate " the triangle ad infinitum ,

As Ive demonstrated, simple visual interpolation is sufficient in the case of the close to 30 minute issue. Or replot using more accurate tides , say 30 minute tides where that is available,

Remember SWL claims better accuracy then the RYA method ,

(a) Only if you blindly apply it, which you are not taught in a class to do.

(b) ONLY , where the assumption applies that the hourly tide DOES rmain constant. In practice and from the way tides are lifted from atlas, or chart diamonds, there are far greater errors in the vectors, then any method corrects for. SWLs method "is not more precise", it is only "more precise" if one applies the axiom " The tide does not change over the hour applied" In reality that statement doesn't hold any validity.

Its merely pixies on the head of a pin stuff

IN practic in the application of multi hour complex tides, gross errors appear that dwarf teh theory. This causes either eyeball corrections or return to hourly plots.

Note that in sinusoidal reversing tides you get more precision then actually is computed as the tidal vector errors get cancelled out. ( its not a good one to use to examine any CTS method).


Theres a reason multi hour CTS are not generally taught outside the RYA, The method relies on underlying error prone data and is subject to massive real life errors.

In reality Ive never seen it used in complex situations, as the ground track ( as we've discovered) can be very hard to predict. ( leaving aside the poster child). Its too risky a method in typical areas where there are big tides and usually big obstructions!

Remember you plan is a plan. I think which ever method you choose to pick a CTS, you will have to adjust for changes in leeway, predicted currents, and even traffic. Keeping this mind, whichever method you choose you will be making corrections along the way. The SWL method plans to get you to point B. If you use the RYA method, you can make adjustments (wags if you like) before you get to D so that you will end up at B and you will most likely have a good idea how you are going to do this before you start out. That said will either method survive first contact with the actual conditions? Depends how the vary from planned conditions!

I for one like the SWL method as it plans to get me to my destination. It only took a few mouse clicks, and less than a minute, to come up with a solution to her last example in OpenCPN. Not appreciably longer than the RYA method. takes. So since the SWL method as a planning tool gets me a plan to where I want to go, is a much better than a planning tool that doesn't?

[goboatingnow]

It assumes the tides behaves AT THE AVERAGE during one hour. But inflating the vector triangle applies the average of THE WHOLE PASSAGE. Order of magnitude (roughly) more error. I can't believe you don't see this. C'mon Dave, don't be vain.. We've all made mistakes and said stupid things on this thread, from which we've all learned so much

In the RYA method each tide vector is in essence inflated by a time. ( Thats whats exactly happening).

SO worse case for the RYA, say in a 6 hour plot , with a 30 minute period left in the last hour, each tide vector is increased by 5 minutes. The RYA would contend that it is quite accurate to say that a 2 kn tide valid for 60 minutes is still valid for 65 minutes.

Thats in my opinion as accurate a statement that saying for example a tide stated at say 14:00 at 2kn is still 2kn at 13:30. In one case the RYA is projecting by 8.3% of the original tide table interpolating, In the other , thats approaching 50%.

Precision must be measured against real life. The method in reality is virtually identical to teh RYA, as in most cases D lies close to B. and the times once is inflating or deflating are often in minutes. Theres simply no real world precision in SWLs version then in the RYA version

[goboatingnow]

Remember you plan is a plan. I think which ever method you choose to pick a CTS, you will have to adjust for changes in leeway, predicted currents, and even traffic. Keeping this mind, whichever method you choose you will be making corrections along the way. The SWL method plans to get you to point B. If you use the RYA method, you can make adjustments (wags if you like) before you get to D so that you will end up at B and you will most likely have a good idea how you are going to do this before you start out. That said will either method survive first contact with the actual conditions? Depends how the vary from planned conditions!

I for one like the SWL method as it plans to get me to my destination. It only took a few mouse clicks, and less than a minute, to come up with a solution to her last example in OpenCPN. Not appreciably longer than the RYA method. takes. So since the SWL method as a planning tool gets me a plan to where I want to go, is a much better than a planning tool that doesn't?

NO the application of each gives almost exactly the same answers , especially where in real life the tide plots come within say 5-15 minutes of teh destination,

each gives you a single CTS and a time to run. IN my answer to her example , visual interpolation gives a CTS of 58.5 and a time to run of approx 2.5. IN reality you probably round it off to 60 ( which was said about 1000 threads ago)

IN the real world the errors in the underlying data and the assumptions in the CTS model are far greater then the difference

Thats even before you add in the factors that affect the boat, including the speed variations , inability to steer better then +- 5 degrees,leeway and wave induced issues.
The base tidal data is not there to support Seaworthys "precision"

[Dockhead]

Originally Posted by goboatingnow

If you want correct precision, that replot ALL the tides with the time interval decreasing towards the greatest precision the tide data is available for. Thats precision.

Not what were talking about.

If we have tide data at a higher resolution, then might as well use it for the whole passage, at the cost of a much increased volume of computation, which may or may not be worthwhile.

But this misses the point that SWL gives a way to a big increase in precision at almost no extra cost, in terms of effort. That can only be good.

Whether it's fool's gold or not is an empirical question, which needs math and/or experiments to solve.

[Dockhead]

Originally Posted by goboatingnow

In the RYA method each tide vector is in essence inflated by a time. ( Thats whats exactly happening).

SO worse case for the RYA, say in a 6 hour plot , with a 30 minute period left in the last hour, each tide vector is increased by 5 minutes. The RYA would contend that it is quite accurate to say that a 2 kn tide valid for 60 minutes is still valid for 65 minutes.

Thats in my opinion as accurate a statement that saying for example a tide stated at say 14:00 at 2kn is still 2kn at 13:30. In one case the RYA is projecting by 8.3% of the original tide table interpolating, In the other , thats approaching 50%.

Precision must be measured against real life. The method in reality is virtually identical to teh RYA, as in most cases D lies close to B. and the times once is inflating or deflating are often in minutes. Theres simply no real world precision in SWLs version then in the RYA version

You will often get a decent answer out of RYA, I admit. It's a decent quick 'n dirty solution. But if the last partial hour is at the bottom or top of a current trend, your slightly inflated collective vectors ( which is nothing other than the extrapolated average), will be WAY out. This is unnecessary.

[goboatingnow]

But this misses the point that SWL gives a way to a big increase in precision at almost no extra cost, in terms of effort. That can only be good.

It does not give any increase in precision to a properly applied RYA method, taught to a class with the overlay of how to use the method and when to apply intelligent interpolation ,( or in fact not to use multi CTS).

IF anything its an ultimate erroneous computation. There simply isn't the precision that Seawothys is claiming to make the method "any better". Its a greater use of garbage in and gargabe out maxim.

Im not disputing the Mathematical correctness, The RYA is equally mathematically correct ( all its triangles add up correctly). The RYA applies the error over a number of tides, Seaworthy applies it to the last,

The underlying data in fact doesnt actually support either of them

[goboatingnow]

But if the last partial hour is at the bottom or top of a current trend, your slightly inflated collective vectors ( which is nothing other than the extrapolated average), will be WAY out. This is unnecessary.

Equally I could say that if the real tide was simply an average of a tide over an hour, yet the tide at say 0-15,15-30,30-45,45-60 was quite different SWL method of apportioning all the corrections to that tide could equally lead to massive errors


in real life Dockhead, D lies quite close to B, and hence the methods converge. ( as does the error). Only in the worst case of exactly half way does the SWL model and the RYA model show the greatest difference.

NOW, to actually determine precision, take a series of real life passages, extract real life tidal data and compare the answers. I think youll see these are really exactly the same methods, in that the circle of error of both will lie INSIDE the circle of error for the data. You have to get away from stylised data that SWL uses , thats why all RYA examples, require you to extract all the data from the word go.

IN a class of 50 students all doing CTS calculations on the same problem, the RYA allows a small range of CTS to be right ( because of tide interpolation). Its simply not anywhere as precise as being presented.

[jackdale]

Originally Posted by Dockhead

It assumes the tides behaves AT THE AVERAGE during one hour. But inflating the vector triangle applies the average of THE WHOLE PASSAGE. Order of magnitude (roughly) more error. I can't believe you don't see this. C'mon Dave, don't be vain.. We've all made mistakes and said stupid things on this thread, from which we've all learned so much

Current tables (the source of the data) publish the data for the maximums and turns. Current graphs that use current diamonds provide the data in hourly intervals for the times calculated. These are not the means of the current over the hour. Electronic tide tables that use sinusoidal waves show the current at a specific time.

Using official tide and current tables, privately published one and the ones on my garmin gps, netbook (openCPN) and Android phone (Navionics) all give me different answers. I expect that RYA uses official publications.

And none of these account well for river outflow or meteorological phenomena.

I live with very large tides (18 foot range) and currents (16.5 knot maximums).

[Dockhead]

Originally Posted by goboatingnow

Equally I could say that if the real tide was simply an average of a tide over an hour, yet the tide at say 0-15,15-30,30-45,45-60 was quite different SWL method of apportioning all the corrections to that tide could equally lead to massive errors


in real life Dockhead, D lies quite close to B, and hence the methods converge. ( as does the error). Only in the worst case of exactly half way does the SWL model and the RYA model show the greatest difference.

NOW, to actually determine precision, take a series of real life passages, extract real life tidal data and compare the answers. I think youll see these are really exactly the same methods, in that the circle of error of both will lie INSIDE the circle of error for the data. You have to get away from stylised data that SWL uses , thats why all RYA examples, require you to extract all the data from the word go.

IN a class of 50 students all doing CTS calculations on the same problem, the RYA allows a small range of CTS to be right ( because of tide interpolation). Its simply not anywhere as precise as being presented.

If the rate of change is constant, then averaging an hour compared to averaging multiple hours is a proportionate difference in error . . . C'mon Dave, that's math even simple enough for me

[jackdale]

Originally Posted by Dockhead

You will often get a decent answer out of RYA, I admit. It's a decent quick 'n dirty solution. But if the last partial hour is at the bottom or top of a current trend, your slightly inflated collective vectors ( which is nothing other than the extrapolated average), will be WAY out. This is unnecessary.

This is part of the passage planning process. During the execution of the plan you had better be getting a fix and then recalculating everything. I want to be able to do that quickly. Any who followed either method for 4 hours without monitoring progress is a fool.

For coastal cruising I want a fix every 30 minutes, for near coastal every hour. For water every watch change.

[goboatingnow]

for example , redo the given example of 3,2,0.5 to 0.5,3,2
compare the methods the CTSs are almost identical ( with +- degree).

see is just a pick your poision

[Andrew Troup]

Originally Posted by Dockhead

There is no mathematical operation which will give you an angle which will give you a course to steer which does not involve, directly or indirectly, implicitly or explicitly, solving a triangle. You can't solve a vector without triangles, because vectors ARE triangles.

Dockhead

You make an assertion without any attempt at explaining it, so all I can do is explain why I believe your assertion is wrong


I'm going to talk about 2D vectors rather than the more general 3D case, given that we're on a sailing, not flying, forum ...

1. 1. Vectors, in relation to navigation, means Euclidean vectors
   2. Euclidean vectors are geometrical entities which have length and direction
      (from here on I will drop the "Euclidean" qualifier)
   3. Their direction is 'signed', meaning that start point is at a specified end
   4. Their direction can be expressed as an angle with respect to a datum vector.
      By convention, this is a vector of unspecified length, whose end point lies True North of its start point
   5. Vectors can be added and subtracted
   6. A simple way to add several vectors is to notionally move them and attach them, head to tail.
   7. The result of such an addition is a single vector, starting at the start point of the first vector and ending at the end point of the last vector.
   8. Subtraction is carried out by reversing the "sign" of one or more vectors, which requires swapping the start and end points.
   9. No other operations are indispensable to the 'solution' (addition and subtraction) of vectors

OK: This competes head-on with your assertion that vectors are triangles.


Please tell me which of the above steps is wrong, which step relies on one or more triangles, or how the steps fail to proceed to the last point.

- - - - -

Navigators are taught to think of vectors as triangles.

Presumably this is because triangles are a useful way to represent vectors. One of several useful ways.

If you truly believe that representations are the same as things, then I would be happy to purchase your sailing vessel. I will pay any price.

Please let me know how many banknotes to photocopy.

[goboatingnow]

If the rate of change is constant, then averaging an hour compared to averaging multiple hours is a proportionate difference in error . . . C'mon Dave, that's math even simple enough for me

If you have an average value that happens to be constructed from large swings, and your proportionality lands you in that time quantum( of an hour) then that produces a large error.since you are assuming an average for that time quantum that in reality may be different.

What you end up here DockHead is an argument over the validity of the tidal data, in Seaworthy case the claimed precision is based on a premise, and we know that premise has no more precision in it than any other.

again look at the example I gave, the two methods converge and in essence both are based on the same underlying errors.

SWL isnt a new method or even a better one , its merely basing a calculation on a "particular" assumption. IN real life that assumption isnt true.

Not only is it foolish to calulate precision where none exists, its actually wrong and leads to blind trust in "answers" , rather like the GPS issue we have discussed