Re: Kriging the data with geological faults

["Edzer J. Pebesma" <e.pebesma@geog.uu.nl>]

Hi Konstantin,

Steve Joyce (steve.joyce@resgeom.slu.se), the writer of the quadtree
routines in search.c, once implemented exactly what you want using gstat
and the POLYGON stuff. However, he never returned it to me, and we never
got to merging it into the main gstat distribution. Maybe, if he's 
willing to help you (sorry, Steve!) we can do that after all.
--
Edzer

Konstantin Malakhanov wrote:
> 
> Hi all GSTAT users,
> 
> I'd like to use GSTAT to krige a base height of geological layers. The
> problem is, the layers have geological faults (not closed ones). So,
> to calculate a value at the point (x,y) one has to take into account
> only the sample points at the same side with (x,y) of all faults. Like
> that (x- sample points, X-sample points to use, o - point to
> interpolate at):
> 
> ---------------------------------------|
> |......................................|
> |...x./................................|
> |..../..............|..................|
> |.../...............|...x..............|
> |../.........X......|..................|
> |...................|..................|
> |...................|..................|
> |......X.........../...................|
> |................./....................|
> |................/........x............|
> |..........X..../...x..................|
> |............../.......................|
> |...o........./........................|
> |......................................|
> ---------------------------------------
> 
> Now, is there a possibility to do so with GSTAT? I was thinking about
> using somehow stratified kriging... There was also some code related
> to POLYGONs (not used?? not documented??)...

(stratification won't work here, since the faults don't split the area)

> 
> Another question: if I would code that functionality, what be the best
> place to put it in? (In low-levels routines in search.c or in higher
> ones in select.c?). I think about it like putting the distances
> between  o-x to a very high value, so they will be sorted out.
> 
> Thanks,
> Konstantin Malakhanov,
> Institut for Hydraulic Engineering & Groundwater Mgmt.,
> RWTH Aachen, Germany
> E-Mail: kosta@iwwnt.iwwl.rwth-aachen.de          Tel. 0241/807343