Get intersects of indexed grouped polygons between 2 SpatialPolygonDataframes in R

Question

I am testing 2 different methods (m2b & hmm) for classifying the movement behaviour of penguins to ultimately classify feeding behaviour from GPS points and from these create kernel polygons to identify foraging areas. I want to check the overlap (area) between the foraging kernels from the 2 different methods. I have 1 SpatialPolygonsDataframe for each method which has a number of foraging kernels for each animal ID. Ultimately for each ID (which can be > 1 polygon) I want to calculate the overlap area between the 2 SpatialPolygonsDataframes. I have tried using rgeos::gIntersection:

o1 <- gIntersection(ud50_m2b,ud50_hmm,byid = T, id = 'ID')

but it returns the following error:

Error in RGEOSBinTopoFunc(spgeom1, spgeom2, byid, id, drop_lower_td, unaryUnion_if_byid_false,  :   ids vector has incorrect length of 1 expected length of 35910

My data looks like this:

> str(ud50_m2b)
Formal class 'SpatialPolygonsDataFrame' [package "sp"] with 5 slots
  ..

@ data       :'data.frame': 2 obs. of  4 variables:
  .. ..$ ID    : Factor w/ 2 levels "CB1201","CB1203": 1 2
  .. ..$ area  : num [1:2] 1971 1178
  .. ..$ survey: Factor w/ 15 levels "","April2012BI",..: 2 2
  .. ..$ n     : int [1:2] 3024 4326
  ..

@ polygons   :List of 2
  .. ..$ :Formal class 'Polygons' [package "sp"] with 5 slots
  .. .. .. ..@ Polygons :List of 3
  .. .. .. .. ..$ :Formal class 'Polygon' [package "sp"] with 5 slots
  .. .. .. .. .. .. ..@ labpt  : num [1:2] 435288 6250135
  .. .. .. .. .. .. ..@ area   : num 5857355
  .. .. .. .. .. .. ..@ hole   : logi FALSE
  .. .. .. .. .. .. ..@ ringDir: int 1
  .. .. .. .. .. .. ..@ coords : num [1:30, 1:2] 434056 433968 433882 433922 434056 ...
  .. .. .. .. ..$ :Formal class 'Polygon' [package "sp"] with 5 slots
  .. .. .. .. .. .. ..@ labpt  : num [1:2] 442476 6251628
  .. .. .. .. .. .. ..@ area   : num 13697898
  .. .. .. .. .. .. ..@ hole   : logi FALSE
  .. .. .. .. .. .. ..@ ringDir: int 1
  .. .. .. .. .. .. ..@ coords : num [1:64, 1:2] 439588 439576 439588 439740 439983 ...
  .. .. .. .. ..$ :Formal class 'Polygon' [package "sp"] with 5 slots
  .. .. .. .. .. .. ..@ labpt  : num [1:2] 442305 6247942
  .. .. .. .. .. .. ..@ area   : num 156406
  .. .. .. .. .. .. ..@ hole   : logi FALSE
  .. .. .. .. .. .. ..@ ringDir: int 1
  .. .. .. .. .. .. ..@ coords : num [1:8, 1:2] 442353 442112 442185 442353 442474 ...
  .. .. .. ..@ plotOrder: int [1:3] 2 1 3
  .. .. .. ..@ labpt    : num [1:2] 442476 6251628
  .. .. .. ..@ ID       : chr "CB1201"
  .. .. .. ..@ area     : num 19711659

I think there is an indexing issue here in that the gIntersection function does not register the same IDs for multiple polygons in each SpatialPolygonsDataframe. Is there a trick to getting the overall intersection area for each ID between the 2 SpatialPolygonsDataframe?

Answer 1

This is a bit long-winded but it works:

    library(rgeos)

overlap<-function(data1,data2){

o1 <- gIntersection(data1,data2,byid = T)

odf1 <- data.frame(id = getSpPPolygonsIDSlots(o1), intersect_area = gArea(o1, byid=TRUE))
odfmeth1 <- data.frame(id = getSpPPolygonsIDSlots(data1), aream1 = gArea(data1, byid=TRUE))
odfmeth2 <- data.frame(id = getSpPPolygonsIDSlots(data2), aream2 = gArea(data2, byid=TRUE))

odf <- odf1 %>% separate(id, into=c("id1","id2"), sep = ' ') %>%
  mutate(id = id1) %>%
  filter(id1==id2) %>%
  left_join(odfmeth1,by='id') %>%
  left_join(odfmeth2,by='id') %>%
  mutate(propm1 = intersect_area/aream1,
         propm2 = intersect_area/aream2) %>%
  dplyr::select(-id1,-id2)

}

ov_ud50_m2b_hmm<- overlap(ud50_m2b,ud50_hmm)
mean(ov_ud50_m2b_hmm$propm1)
sd(ov_ud50_m2b_hmm$propm1)