epix-project / sptools Goto Github PK

The function make_grid() returns a SpatialPointsDataFrame object with the coordinates as variables. It doesn't make much sense and contains the coordinates informations twice. It would make more sense if make_grid() returned a SpatialPoints object instead.

That should probably be a method in order to deal with SpatialPoints, SpatialPointsDataFrame, SpatialPolygons, SpatialPolygonsDataFrame, etc...

Can you check the new function rescale_raster please?
Including trying the run the example?
(I know... I should have made a branch for that... Next time!)

The function gadm download the last version of gadm file, is it possible to remove it after reading the file?

Would it be interesting to add a buffer arguments in the split_on_poly and crop_on_poly functions, to be able to select all the value of the raster even the cells that overlap with the polygon?

As the package imhen has been updated and the output stations is now an sf class object. The examples using this package are not working anymore and need to be updated.

For resample_from_grid, split_on_poly and add_from_raster, what happens if the Raster* and the Spatial* objects do not superpose? Does it returns an error? (I guess so). Should we include a test on that in these functions?

Here is a demonstration of the bug. For that, we need the following packages:

# devtools::install_github("choisy/srtmVN")
# devtools::install_github("choisy/worldpopVN")
library(magrittr)
library(sp)

The data we need:

country <- sptools::gadm("vietnam", "sp", 0)
elevation <- srtmVN::getsrtm()
popdensity <- worldpopVN::getpop()

Let's make a grid:

grid <- country %>%
  sptools::make_grid(100) %>%
  sptools::add_from_raster(elevation, "elevation")

Let's make a second version of this grid, where we remove the variable longitude:

grid2 <- grid
grid2@data[, "longitude"] <- NULL

So, the only difference between grid and grid2 is that grid contains the variable longitude that grid2 does not have. We can check that the coordinates are the same:

identical(coordinates(grid), coordinates(grid2))

The bug is due to the fact that

sptools::resample_from_grid(popdensity, grid)

works just fine whereas

sptools::resample_from_grid(popdensity, grid2)

fails with the following message:

Error in .intersectExtent(x, y, validate = TRUE) : 
  Objects do not intersect 

Can we use the add_from_raster() function on a SpatialPoints object instead of a SpatialPointsDataFrame object? If no, should we update this function so that it can deal with SpatialPoints objects as well as with SpatialPointsDataFrame objects? (I think we should).

Let's consider the climatic stations in the province of Hanoi:

library(sp)
provinces <- sptools::gadm("vietnam", "sp", 1)
hanoi <- subset(provinces, VARNAME_1 == "Ha Noi")
stations <- as(imhen::stations, "Spatial")

From the plot, we can see that there are 3 stations in the province of Hanoi:

plot(hanoi)
plot(stations, add = TRUE)

Yet, the function points_in_polygons fails to return 3 points:

length(sptools::points_in_polygon(stations, hanoi))