Subject: Re: how to determine distance from line? [BoundaryPoint] 1542
Date: Sep 10, 2004 @ 07:26
Author: aletheiak ("aletheiak" <aletheiak@...>)
Prev Post in Topic Next [All Posts]
Prev Post in Time Next
>
> "...daves report roughly midway down the page
>
> <http://tinyurl.com/6zgwc> "
>
> Ah. A key message I apparently overlooked while out of town.
> That answers some questions I asked in a later message.
>
> "...limit our overall degree of certainty
> to 2 decimal places..."
> Yep, at best 2 places.
> My programs use about 12 digits total
> for deg min sec and other calculations.
> BUT, it can't make the output better
> than the output.
>
> "..the submillimetric exactitude of your solution
> here below seems to indicate that you dont agree..."
> I believe we are on the same page and same song.
> One way to try to preserve the 2 places input
> is to do the arithmetic to at least one more
> and then round to 2 at the end.
> The program just shows all the digits.
>
> > But, how to find the north-south distance from
> > a 3rd point to an intermediate point on the line
> > isn't obvious.
>
> "ok but i still think it is obvious
> at least within the above stated parameters of exactitude
> it seems entirely reducible to a simple ratio computation"
> True for the short distances in this problem,
> but for the general case where the points
> are a lot further apart plane trig stops working
> and spherical trig has to be used, as in the formula
> (which works for short and long if one is careful
> to account for the special cases).
>
> > ==========================
> >
> > RCMcC
> > Angles and Distance greatly distorted.
>
> "what do you mean specifically by the above comment?"
>
> Just that it's hard to draw a triangle
> that's 3212 meters long and only 0.08 meters tall.
>
> >
> > [194] -- 3212 m -- I -- [195]
> > \ | /
> > 2946 m 267.048 m
> > \ /
> > [LSAW 5]
> >
> > ==========================================
>
> "> Given (all NAD-27 converted to NAD-83 with CORPSCON )
> >
> > Monument #194 = Point 1 = Lat1/Long1 = N 48.999056º W
> 117.072683º
> > Monument #195 = Point 2 = Lat2/Long2 = N 48.999215º W
> 117.028771º
> > LSAW#5 plaque = Point 3 = Lat3/Long3 = N 48.999197º W
> 117.032421º
>
> "but i think your 6th digit may be questionable
> for points 1 & 2"
>
> Maybe so. I used all the NAD-27 digits
> to feed into the standard CORPSCON translator to NAD-83
> and then rounded that output off to 3 digits
> to try to keep as much as possible until after
> all the number crunching.
> The final result after all calculations for Point I
> on the boundary is only good to 2 digits - at most.
> Then one should also round points 1 and 2.
>
> "...crediting such exactitude as you are delivering..."
> The math is just manipulating a model of physical reality.
> When physical reality is different from all the
> digits put out by the computer,
> physical reality rules
> and you try to correct the model.
>
> "...someone who understands all this ..."
> Gotcha fooled. :)
> I'm definitely not a pro at this stuff,
> just someone who has fiddled with it in spare
> time for a long time trying to keep the models
> (borrowed from the _real_ pros) as accurate as possible
> when crunching the numbers through a computer.
>
> I've studied a lot of sloppy programs
> where someone encoded a formula from a textbook
> into a computer program and had no idea
> what its limitations were and how the limitations
> of the computer arithmetic affected it.
> The GUIs looked great, though.
>
> A few years ago I was reading a program
> where I found that the programmer had
> twice typed in the value of pi as
> 3.141592818
> where the three digits "818" after the "2" are wrong.
> Fortunately, in his particular case it made
> no real difference, but if he were trying
> to calculate distances down to inches,
> as we are attempting here - no good.
> I guessed that he may have obtained
> the wrong value of pi from the MS Windows calculator.
>
> We are going out of town tomorrow and it may be
> Monday before I get to check in again if I miss
> in the morning.
>
> Have fun.
>
>
> Cheers, 73,
>
> Ron McC.
> w2iol@a...
>
> Ronald C. McConnell, PhD
>
> WGS-84: N 40º 46' 57.6" +/-0.1"
> W 74º 41' 22.1" +/-0.1"
> FN20ps.77GU31 +/-
> V +5058.3438 H +1504.2531
>
> http://home.earthlink.net/~rcmcc
>
> There are 10 kinds of people.
> Those who understand binary arithmetic
> and those who do not.