Calculating chi maps from simulated corridors #23
Description
Dear trajectories developers (and users),
First of all, congrats for the package. The general ideia is simple and great, and I found this chi square maps particularly interesting, since I haven't seen that in a spatial form before.
I have simulated ecological corridors, for a given species, in two different land use scenarios. I am willing to compare where their simulated tracks differ using chi maps. The case is a little different from the vignette, since we do not have individuals, but simulated paths in different scenarios. Also, we don't have time (it is implicit), but I thought about considering each scenario as times far away from each other, so that they may be compared.
Do you think this is possible?
This is what I did:
Each simulated corridor is a shapefile. I have 1000 simulated corridors for each land use scenario.
First I resampled each of them as equally distance points along the paths, so that each corridor may be considered as sequences of points and the spatial point processes may be applied.
# Sample of code, after importing corridors
nn <- round(gLength(corridor) / 300)
pts <- spsample(corridor, n = nn, type = "regular")
# Here I added time in a random way, but I am not sure how should I have done that
t0 <- as.POSIXct(as.Date('2019-01-02 00:00', tz = 'UTC')) + 100*j # j is the index for corridor number
df <- data.frame(scenario = rep(1, nn), corr = rep(j, nn), t = t0 + 1:nn)
spdf <- SpatialPointsDataFrame(pts, df)
# This piece of code was included inside a loop over all corridors
I did not know how to include time, so I considered each location as 1s for the previous, and each corridor stating in different moments in time, to avoid overlap.
Now, I had a list with all 1000 corridors for a single land use scenario: spdf.l.landuse1.
What I did was to transform each corridor into a STIDF and then into a Track:
track.landuse1 <- list()
for(j in 1:length(spdf.l.landuse1)) {
print(j)
sti <- STIDF(sp = SpatialPoints(spdf.l.landuse1[[j]]@coords, proj4string = spdf.l.landuse1[[j]]@proj4string),
time = spdf.l.landuse1[[j]]$t, data = spdf.l.landuse1[[j]]@data[,1:2])
track.landuse1[[j]] <- Track(sti)
}
Finally, I transformed that into a Tracks object:
tracks.landuse1 <- Tracks(track.landuse1)
I could successfully plot that (below), and it is now consider as a series of 1000 tracks of the same object. However, I guess that it would be better to make a TracksCollection directly, considering each corridor as a Track of a different individual. This way, we could consider all tracks of a given lan use scenario as simultaneous movements of different individuals, and then compare them with the other land use scenario by choosing a very different moment in time for the second land use scenario.
However, I couldn't transform the tracks directly into a TracksCollection object.
Still, I tried to move on.
All the analyses I tried to perform then (movement smoothing and chi maps), presented the same error: Here is the code and the error:
# movement smoothing
b <- Track.idw(tracks.landuse1, timestamp = '20 days')
plot(b, main="", ribwid=0.04, ribsep=0.02, cex.axis=1.5)
# chi maps
ch <- chimaps(tracks.landuse1, timestamp = "20 mins", rank = 1)
# Error:
Error in if (!ivs[i] == 0 && !ivs[i] == nrow(X)) { :
missing value where TRUE/FALSE needed
Do you know what it may mean?