[GRASSLIST:189] RE: How to report length totals for all lines mat ching specific ID

[Hamish]

> > I have a vector of a coastline; the coast is segmented into
> > different coastal types, according to unique IDs:
..
> > I successfully popluated the SEG_LENGTH column with the length of
> > each segment using v.to.db:
..
> Note coastline length is a fractal problem and answers are only useful
> when compared against the same dataset, e.g. as a percentage of the
> whole. Even then trusting it is a bit dubious (same digitizer at the
> same level of fatigue, etc).
> 
> see-
> http://mathworld.wolfram.com/CoastlineParadox.html
> http://www.vanderbilt.edu/AnS/psychology/cogsci/chaos/workshop/Fractals.html
> http://en.wikipedia.org/wiki/How_Long_Is_the_Coast_of_Britain%3F_Statistical_Self-Similarity_and_Fractional_Dimension
..
> Thanks for bringing up that point. Fortunately, the coastline I'm
> reporting lengths on is a 'final' product released by another gov't
> department, and so is likely to be the authoritative source for quite
> some time (presumably). Interesting links you sent.

I should also mention that besides the choices of the digitizer hand &
processing, fractal depth will also be a function of the local geology 
and wave exposure.

Clearly stating which dataset the length data was derived from helps,
but IMO "coastline length" a measure that policy makers should be
steered away from at every opportunity. It's sexy but bogus.



Hamish