[GRASS-user] Working with transects

[Dylan Beaudette]

On Monday 14 January 2008, Maris Nartiss wrote:
> Hello GRASSers!
>
> I have transect data as points and I need to identify slope
> break-points along transect. If anyone has worked with similar
> problem, can offer some solution or give a peace of advice - I would
> be thankfull.

Hi,

>
> More detailed description.
> I have height measurements made in form of transects (see attached
> sample). Those transects are almost straight and almost perpendicular
> to ancient shorelines (strandlines). My task is simple - I have to
> create profiles in order to locate ancient shoreline exact position
> (they should show up as specific break points on profile).
> Currently I have imported data as vector points (v.in.ogr), created
> convex hull (v.hull), created buffer around convex hull (v.buffer),
> converted result to raster mask (v.to.rast) and then interpolated
> surface from loaded points with various methods (v.surf.rst,
> v.surf.idw, v.surf.bspline).

It looks like you are making good progress with the GRASS tools, and following 
conventions generally used for interpolation of surfaces. Note that cells 
outside of the convex hull are much less meaningfull than those nearby your 
starting points.

> Interpolation results differ significantly (see attached sample of
> same profile line values on different interpolated surfaces). I know,
> that I can tweak v.surf.rst parameters to get better(?) results, still
> I have not yet worked with such spatial distribution data and thus
> have no idea how to do that.

RST is generally the best, unless it is producing results which are too 
smooth: i.e. when there are discontinuities or other 'breaking' features. It 
does not sound like this is the case for your project.

> List of current problems/shortcomings:
> * how to extract vector point values along profile line or get vector
> point attribute data for nearest points to profile line (vector line
> in general)? v.profile/v.transect?

Hmmm... not sure that I understand the question. Are you trying to do some 
kind of 'spatial join' of attribute data, where the attributes of some vector 
which is (closest?) to another vector are selected? If this is the case 
PostGIS might be of interest- spatial + attribute queries in SQL. Otherwise 
you might look into r.cost / v.distance / v.to.db ... ?


> * which of the GRASS offered data interpolation methods is the best to
> interpolate data on transect?

This is a good question. I *think* that the methods in GRASS are designed for 
interpolation of 2D or 3D _surfaces_. If you are interested in 1D 
interpolation I would suggest using R. Here is a link to some GRASS-R raster 
profile stuff:
http://casoilresource.lawr.ucdavis.edu/drupal/node/375

In the case of 1D interpolation, check out the following commands:
?spline
?lowess
library(splines)
?bs
?ns


> * any other tools, that could suit better for this task?
>
> Thanks for reading this and sorry for my clumsy language,
> Maris.


Quick question- are you looking for terraces along these transects?


cheers,
-- 
Dylan Beaudette
Soil Resource Laboratory
http://casoilresource.lawr.ucdavis.edu/
University of California at Davis
530.754.7341