sdmtools

A set of helper functions to facilitate species distribution modelling.

Installation

You can install sdmtools with:

install.packages(
"sdmtools",
repos = "https://idem-lab.r-universe.dev"
)

Data

library(sdmtools)

raster_to_terra — an annotated equivalence table of functions from the raster and terra. First 5 lines:

global_regions — a tibble showing the WHO region, UN region, and continent for for 249 countries and country-like things. First 5 lines:

Data-generating functions

The package terra is fiddly about storing its spat... objects in packages, so we chose to generate example spatial data on-demand using functions, rather than storing it.

example_raster — an example spatRaster.

library(terra)
#> terra 1.7.79

r <- example_raster()
r
#> class       : SpatRaster 
#> dimensions  : 10, 10, 1  (nrow, ncol, nlyr)
#> resolution  : 1, 1  (x, y)
#> extent      : 0, 10, 0, 10  (xmin, xmax, ymin, ymax)
#> coord. ref. :  
#> source(s)   : memory
#> name        :   example 
#> min value   : 0.0627102 
#> max value   : 7.3352526

plot(r)

example_vector — an example spatVector.

library(terra)
v <- example_vector()
v
#>  class       : SpatVector 
#>  geometry    : points 
#>  dimensions  : 10, 0  (geometries, attributes)
#>  extent      : 0.2293562, 8.00672, 1.375653, 8.951683  (xmin, xmax, ymin, ymax)
#>  coord. ref. :

plot(v)

make_africa_mask — makes a mask layer of Africa based on shapefiles from malariaAtlas::getShp. Can produce either a SpatRaster or SpatVector.

library(terra)
africa_mask <- make_africa_mask(type = "vector")
#> Loading ISO 19139 XML schemas...
#> Loading ISO 19115 codelists...
#> Please Note: Because you did not provide a version, by default the version being used is 202403 (This is the most recent version of admin unit shape data. To see other version options use function listShpVersions)
#> although coordinates are longitude/latitude, st_union assumes that they are
#> planar
#> Warning: [crs<-] not all geometries were transferred, use svc for a geometry
#> collection

plot(africa_mask)

Function examples

rastpointplot — simple utility to plot a raster with points over it.

rastpointplot(r,v)

source_R — source all R files in a target directory

source_R("/Users/frankenstein/project/R") # do not run

import_rasts — import all rasters from a directory into a single object

rasters <- import_rasts("/data/grids/covariates") # do not run

split_rast — split a raster.

r <- example_raster()

s <- split_rast(r, grain = 2)

s
#> [[1]]
#> class       : SpatRaster 
#> dimensions  : 5, 5, 1  (nrow, ncol, nlyr)
#> resolution  : 1, 1  (x, y)
#> extent      : 0, 5, 0, 5  (xmin, xmax, ymin, ymax)
#> coord. ref. :  
#> source(s)   : memory
#> name        :   example 
#> min value   : 0.1587361 
#> max value   : 7.3352526 
#> 
#> [[2]]
#> class       : SpatRaster 
#> dimensions  : 5, 5, 1  (nrow, ncol, nlyr)
#> resolution  : 1, 1  (x, y)
#> extent      : 0, 5, 5, 10  (xmin, xmax, ymin, ymax)
#> coord. ref. :  
#> source(s)   : memory
#> name        :   example 
#> min value   : 0.1028045 
#> max value   : 4.0001839 
#> 
#> [[3]]
#> class       : SpatRaster 
#> dimensions  : 5, 5, 1  (nrow, ncol, nlyr)
#> resolution  : 1, 1  (x, y)
#> extent      : 5, 10, 0, 5  (xmin, xmax, ymin, ymax)
#> coord. ref. :  
#> source(s)   : memory
#> name        :    example 
#> min value   : 0.09802478 
#> max value   : 3.23820739 
#> 
#> [[4]]
#> class       : SpatRaster 
#> dimensions  : 5, 5, 1  (nrow, ncol, nlyr)
#> resolution  : 1, 1  (x, y)
#> extent      : 5, 10, 5, 10  (xmin, xmax, ymin, ymax)
#> coord. ref. :  
#> source(s)   : memory
#> name        :   example 
#> min value   : 0.0627102 
#> max value   : 5.7145289

ps <- lapply(
  s,
  FUN = extend,
  y = r
) |>
  rast()

c(
  r,
  ps
) |>
  plot()

Functions for a species distribution modelling workflow

We have some covariate layers: cov1 and cov2

library(terra)

cov1 <- example_raster(
  seed = -44,
  layername = "cov1"
)
cov2 <- example_raster(
  seed = 15.3,
  layername = "cov2"
)

covs <- c(cov1, cov2)

std_rast — standardise a spatRaster by transforming it to have a range of 0—1

cov1_st <- std_rast(cov1)

plot(cov1_st)

We have some presences and absences

presences <- example_vector(seed = 68) %>%
  as.data.frame(geom = "xy")
absences <- example_vector(seed = 9.6) %>%
  as.data.frame(geom = "xy")

presences
#>           x        y
#> 1  9.244899 5.033042
#> 2  6.612025 1.559797
#> 3  4.024099 8.750261
#> 4  6.370063 4.438317
#> 5  3.526324 6.598762
#> 6  7.476441 7.754586
#> 7  7.175489 8.123659
#> 8  1.935898 5.082858
#> 9  3.331217 7.974853
#> 10 1.365547 5.741829

extract_covariates — extract covariate values from spatRaster or raster layers for a given set of points

Pass in either presences and absences as a data.frame or tibble of with , or presences_and_absences as a single data frame points with a presence or ID column(s)

sdm_data <- extract_covariates(
  covariates = covs,
  presences = presences,
  absences = absences
)

We can then make a spatial prediction of our model using predict_sdm and write and read it out in a single step with writereadrast, and write it to a temporary file with temptif:

# first we make a simple model, using data from above
m <- glm(presence ~ cov1 + cov2, data = sdm_data)

prediction_rast <- predict_sdm(m, covs) |>
  writereadrast(filename = temptif())

plot(prediction_rast)