Subject:
Re: [GRASS-user] Length of a political boundary
Date:
Mon, 04 Apr 2011 07:48:02 +0530
From:
ambijat <ambijat@hotmail.com>
To:
Hamish <hamish_b@yahoo.com>
No, Hamish,
This is not the case as I see it. My fundamental method is as follows:
1. in a 3x3 matrix of neighbours analysis I want to fit in one worthy
formula.
2. For every such matrix there are going to be at least 3 non-zero
values and 6 zero values.
3. as we know that c^2 = a^2 + b^2, i would be counting on a = elevation
difference between the two adjacent cells, and b = 90 meters, which is
the resolution of the cell.
4. So, c values for each such set of two adjacent cells can be
calculated, and a sum of all such values which are actually hypotenuse
values would give me more accurate value of the total length of the
boundary.
This is a kind of algorithm I have in mind. Anybody who can translate
into formula?
Thanks!
Ambrish
On 4/4/2011 5:07 AM, Hamish wrote:
> ambijat wrote:
>> I have a typical problem. I have extracted the raster
>> boundary line from SRTM data. each cell contains the
>> elevation value. Idealll\y speaking the total length of a
>> boundary would be the some of the cells x cell resolution
>> (90m). But, this is not true to actual boundary length as
>> there are elevation in mountainous regions.
>>
>> So, I need to add all the hypotenuse values of the cells.
>> How can it be done any idea ?
> you can do lots of refinement, but fundamentally it's a fractal problem
> with no real answer,
> see
> [http://en.wikipedia.org/wiki/How_Long_Is_the_Coast_of_Britain%3F_Statistical_Self-Similarity_and_Fractional_Dimension](http://en.wikipedia.org/wiki/How_Long_Is_the_Coast_of_Britain%3F_Statistical_Self-Similarity_and_Fractional_Dimension)
>
> the 2D plan-area within a boundary will converge on a single answer though
> as you make dx smaller and smaller. also within a single constant dataset
> you can compare one area to another ("this edge is x% of the total
> boundary").
>
> v.to.rast + v.to.db may give you an answer, but I won't use it.
>
>
> see also the r.surf.area and r.volume modules, which are slightly different
> creatures in the same family of problem.
>
>
> Hamish
>
>
>