[R-sig-Geo] Sf find number of parts in multigeometry

[Michael Sumner]

I see perfectly good answers to this, but can't resist sharing my own
approach.

 I use  gibble::gibble for a summary of parts. See how the multi-part
object is number 4, and has 3 subobjects (subobject will repeat within
object for holes).

library(sf)
x <- read_sf(system.file("gpkg/nc.gpkg", package = "sf"))
gibble::gibble(x)
#    A tibble: 108 x 5
#     nrow  ncol type         subobject object
#    <int> <int> <chr>            <int>  <int>
#1    27     2 MULTIPOLYGON         1      1
#2    26     2 MULTIPOLYGON         1      2
#3    28     2 MULTIPOLYGON         1      3
#4    26     2 MULTIPOLYGON         1      4
#5     7     2 MULTIPOLYGON         2      4
#6     5     2 MULTIPOLYGON         3      4
#7    34     2 MULTIPOLYGON         1      5
#...

(I use that for mapping out set-based operations on geometry data, it
doesn't make a huge amount of sense on its own. I suppose a rapply scheme
could be constructed to pluck out things, but you'd also want path extents
and sizes and so forth for greater control).

But, if the biggest area of each multipolygon is what you want, give each a
unique ID,  use st_cast to POLYGON, group by parent and slice out the
largest.

library(dplyr)

x %>% mutate(ID = row_number()) %>% st_cast("POLYGON") %>%
group_by(ID) %>% arrange(desc(st_area(.))) %>% slice(1) %>% ungroup()

Note that holes within a part might reduce the area logic, but so will
details of the map projection in use and so on. It's helpful to learn the
structure of the underlying geometry lists to craft your own rogue
solutions.

HTH


On Tue, Oct 23, 2018, 13:32 Martin Tomko <tomkom using unimelb.edu.au> wrote:

> I am looking for an equivalent to Postgis ST_NumGeometries
> https://postgis.net/docs/ST_NumGeometries.html
> I have multipolygons in an sf df, where most of them are likely to be
> single part. I want to identify those that are not single part, and
> possibly retain only the largest polygon part, and cast all into Polygon.
> Any advice is appreciated, thanks!
>
> Martin
>
>
>         [[alternative HTML version deleted]]
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo using r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>
-- 
Dr. Michael Sumner
Software and Database Engineer
Australian Antarctic Division
203 Channel Highway
Kingston Tasmania 7050 Australia

	[[alternative HTML version deleted]]