Skip to content

Too many points for deldir()

3 messages · Rolf Turner, Ludwig Hilger

#
Hello everybody,

I am trying to construct Thiessenpolygons from a very large dataset (ca. 1.3
mio points) and get the following error:

thiessen <- deldir(normv$X, normv$Y)
[...]
Fehler: lange Vektoren (Argument 7) nicht unterst?tzt in  .C (~= long
vectors (argument 7) not supported in .C)

I am wondering if this comes from the number of points and if yes, if there
is a way to split the dataset into blocks and putting the result back
together afterwards? I suspect that there will be problems at the
borderlines of blocks!?

Thank you very much and best regards,
Ludwig

R version 3.0.0 (2013-04-03)
Platform: x86_64-w64-mingw32/x64 (64-bit)

locale:
[1] LC_COLLATE=German_Germany.1252  LC_CTYPE=German_Germany.1252   
LC_MONETARY=German_Germany.1252 LC_NUMERIC=C                   
LC_TIME=German_Germany.1252    

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
[1] sp_1.0-9      deldir_0.0-22

loaded via a namespace (and not attached):
[1] grid_3.0.0      lattice_0.20-15



-----
Dipl. Geogr. Ludwig Hilger
Wiss. MA
Lehrstuhl f?r Physische Geographie
Katholische Universit?t Eichst?tt-Ingolstadt
Ostenstra?e 18
85072 Eichst?tt

Tel.: 08421-93-1180 
--
View this message in context: http://r-sig-geo.2731867.n2.nabble.com/Too-many-points-for-deldir-tp7583537.html
Sent from the R-sig-geo mailing list archive at Nabble.com.
2 days later
#
I had a similar query about a month ago, and it appears that the
limitation is intrinsic to R and there's nothing that can be done about
it through modifying deldir.

You are quite right to say that "there will be problems at the
borderlines of blocks".  I would guess that doing such an amalgamation
of triangulations would be a very difficult problem.  It is possible that
there is something about it in the literature; I wouldn't know.

There is another package for doing triangulation on CRAN.  It's called
"tripack".  This package presumably uses a different algorithm from
deldir so it is *possible* that it does not bump up against the same
limitation.  You might want to give it a try.  There is other software
(not written in R) out there for doing triangulation and it is possible
that one of these implementations might alleviate your problem. There
is a "sweepline algorithm" due to Steve Fortune about which I have
seen laudatory comments.

Finally, let me just say that it is mysterious to me what one could
discover from a triangulation/tessellation of 1.3 million points. It seems
to me that any potential information would be lost in a blur of noise.
Why are you doing this?

     cheers,

         Rolf Turner
On 11/05/13 00:05, Ludwig Hilger wrote:
#
Dear Mr. Turner,

first of all: Thank you very much for you comments and hints! I will give
tripack a try but I am also already working on the problem with other
software. I am trying to use Thiessen polygons to assign areas of uniform
change to very high-density Lidar measurement points, so the change distance
measured at a point would be thought of being representative of the change
within its Thiessen polygon, which will be one way of getting to Cut and
Fill-Volumes.

best regards and thank you again,

Ludwig Hilger 



-----
Dipl. Geogr. Ludwig Hilger
Wiss. MA
Lehrstuhl f?r Physische Geographie
Katholische Universit?t Eichst?tt-Ingolstadt
Ostenstra?e 18
85072 Eichst?tt

Tel.: 08421-93-1180 
--
View this message in context: http://r-sig-geo.2731867.n2.nabble.com/Too-many-points-for-deldir-tp7583537p7583558.html
Sent from the R-sig-geo mailing list archive at Nabble.com.