[R-sig-Geo] [R] polygon shapefile from line edge coordinate list

[Roger Bivand]

On Sun, 9 Mar 2008, Murray Richardson wrote:

> Hi Roger
>
> Thanks so much for your help.  I can definitely work with this.  On your
> advice I'm responding via r-sig-geo.
>
> Unfortunately, there is no guarantee that the line segments will come in
> any particular order from the alpha shapes routine.  In fact, there will
> also likely be line segments from  different polys mixed together, so I
> need to write the code so that the polygons are assembled only for
> connected line segments.  I was naive when I started thinking about the
> problem.  As you suggest, this could get messy.

I understand. I might be tempted to use AWK externally to number the 
unique (x,y) pairs, and reorder the text file externally. There is some 
AWK code and shell scripts in the infrastructure in the maps package and 
the two Bell Labs reports referenced there. They treat the short line 
segment spaghetti scenario directly. Maybe that might provide a way 
forward?

>
> As an alternative I may load the segments into a postGIS linestring
> table and polygonize that, all via RODBC/postGIS.  An added benefit is
> that it would clean up dangling segments, which may also exist.
>

This would be more robust, and via ogr2ogr would let you make the 
shapefiles too. I'd prototype in AWK because I know it adequately and 
don't know PostGIS, but if you already know PostGIS, the underlying code 
is certainly stronger.

Please let us know how you get on, others may face the same problem at 
some time.

Roger

> Thanks again
>
> Murray
>
>
>
> Roger Bivand wrote:
>> Murray Richardson <murray.richardson <at> utoronto.ca> writes:
>>
>>
>>> Hello,
>>>
>>> I am looking for advice on a task I am trying to complete.
>>>
>>> I have a 4 column dataframe defining the start and end coordinates of
>>> line edges (from a CGAL alpha shapes function to define concave hulls
>>> from point clusters).  I would like to create polygon shapefiles from
>>> these line edges, presumably creating lines first and then polygons.
>>>
>>
>> How many polygons in each data.frame? If more than one, how are they
>> separated? You note below that there are multiples, are they flagged by
>> for example an NA row, or is there a jump from (endX endY) to the next
>> (startX startY)?
>>
>>
>>> e.g. columns are:
>>>
>>> startX  startY endX endY
>>>
>>> where each row represents start and end coordinates of a line segment.
>>>
>>
>> Are the line segments ordered in sequence. If they are, something like:
>>
>> xy0 <- data.frame(sX=c(1,2,2,1,3,4,4,3), sY=c(1,1,2,2,3,3,4,4),
>>   eX=c(2,2,1,1,4,4,3,3), eY=c(1,2,2,1,3,4,4,3))
>>
>> brks <- logical(nrow(xy0))
>> for (i in 2:nrow(xy0)) brks[i] <- (xy0$sX[i] != xy0$eX[i-1]) &
>>   (xy0$sY[i] != xy0$eY[i-1])
>> cbrks <- cumsum(brks)+1
>> # find the separate polygons
>>
>> xy1 <- split(xy0, cbrks)
>>
>> library(sp)
>> Plist <- vector(mode="list", length=length(xy1))
>> for (i in seq(along=Plist)) {
>>   crds <- rbind(cbind(xy1[[i]]$sX, xy1[[i]]$sY),
>>     c(xy1[[i]]$eX[nrow(xy1[[i]])], xy1[[i]]$eY[nrow(xy1[[i]])]))
>> #   close the polygon
>>   Plist[[i]] <- Polygons(list(Polygon(crds)), ID=names(xy1[i]))
>> }
>> # make a list of Polygons objects with IDs
>>
>> Plist1 <- SpatialPolygons(Plist)
>> plot(Plist1, axes=TRUE)
>> SPDF <- SpatialPolygonsDataFrame(Plist1, data=data.frame(i=names(xy1),
>>   row.names=names(xy1)))
>> # convert this to a SpatialPolygonsDataFrame object for export
>>
>> library(maptools)
>> writePolyShape(SPDF, "chulls")
>>
>> If the segments are not in "join up the dots" sequence, it will be
>> (even) more messy. Please also consider whether the R-sig-geo list
>> might not be more relevant.
>>
>> Roger
>>
>>
>>> I am new to R spatial packages and I am not sure which one is best
>>> suited to this task. There seems to be a lot of options, which is great
>>> but hard to know which one to start with.  Any suggestions on the best
>>> way to proceed with this?  One further challenge is that the list
>>> includes line segments defining multiple polygons.
>>>
>>> Thanks for any advice.
>>>
>>> Murray
>>>
>>> ______________________________________________
>>> R-help <at> r-project.org mailing list
>>> https://stat.ethz.ch/mailman/listinfo/r-help
>>> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
>>> and provide commented, minimal, self-contained, reproducible code.
>>>
>>>
>>>
>>
>> ______________________________________________
>> R-help at r-project.org mailing list
>> https://stat.ethz.ch/mailman/listinfo/r-help
>> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
>> and provide commented, minimal, self-contained, reproducible code.
>>
>>
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>

-- 
Roger Bivand
Economic Geography Section, Department of Economics, Norwegian School of
Economics and Business Administration, Helleveien 30, N-5045 Bergen,
Norway. voice: +47 55 95 93 55; fax +47 55 95 95 43
e-mail: Roger.Bivand at nhh.no