Migrated R-spatial dependencies

* Replaced `raster`, `rgeos`, and `sp` packages in Imports with `terra` and `sf` because of imminent package retirements * Removed `maptools` from Suggests (replaced with new internal function `as.im.SpatRaster()`) * Thank you, [Roger Bivand](https://github.com/rsbivand), for the notice. Relates to [ndi Issue #3](#3) * Note: `raster` is a dependency of `RStoolbox` (at present) which is used in the vignette * Updated test, examples, vignette, and documentation throughout * Added GitHub R-CMD-check
lance-waller-lab · Dec 16, 2022 · 8bc4025 · 8bc4025
1 parent f305c1a
commit 8bc4025
Show file tree

Hide file tree

Showing 34 changed files with 585 additions and 492 deletions.
diff --git a/.Rbuildignore b/.Rbuildignore
@@ -7,3 +7,4 @@
 ^CRAN-RELEASE$
 ^CRAN-SUBMISSION$
 ^man/figures
+^\.github$
diff --git a/.github/.gitignore b/.github/.gitignore
@@ -0,0 +1 @@
+*.html
diff --git a/.github/workflows/R-CMD-check.yaml b/.github/workflows/R-CMD-check.yaml
@@ -0,0 +1,49 @@
+# Workflow derived from https://github.com/r-lib/actions/tree/v2/examples
+# Need help debugging build failures? Start at https://github.com/r-lib/actions#where-to-find-help
+on:
+  push:
+    branches: [main, master]
+  pull_request:
+    branches: [main, master]
+
+name: R-CMD-check
+
+jobs:
+  R-CMD-check:
+    runs-on: ${{ matrix.config.os }}
+
+    name: ${{ matrix.config.os }} (${{ matrix.config.r }})
+
+    strategy:
+      fail-fast: false
+      matrix:
+        config:
+          - {os: macos-latest,   r: 'release'}
+          - {os: windows-latest, r: 'release'}
+          - {os: ubuntu-latest,   r: 'devel', http-user-agent: 'release'}
+          - {os: ubuntu-latest,   r: 'release'}
+          - {os: ubuntu-latest,   r: 'oldrel-1'}
+
+    env:
+      GITHUB_PAT: ${{ secrets.GITHUB_TOKEN }}
+      R_KEEP_PKG_SOURCE: yes
+
+    steps:
+      - uses: actions/checkout@v3
+
+      - uses: r-lib/actions/setup-pandoc@v2
+
+      - uses: r-lib/actions/setup-r@v2
+        with:
+          r-version: ${{ matrix.config.r }}
+          http-user-agent: ${{ matrix.config.http-user-agent }}
+          use-public-rspm: true
+
+      - uses: r-lib/actions/setup-r-dependencies@v2
+        with:
+          extra-packages: any::rcmdcheck
+          needs: check
+
+      - uses: r-lib/actions/check-r-package@v2
+        with:
+          upload-snapshots: true
diff --git a/DESCRIPTION b/DESCRIPTION
@@ -1,8 +1,8 @@
 Package: envi
 Type: Package
 Title: Environmental Interpolation using Spatial Kernel Density Estimation
-Version: 0.1.15
-Date: 2022-08-30
+Version: 0.1.16.9000
+Date: 2022-12-16
 Authors@R: 
     c(person(given = "Ian D.",
              family = "Buller",
@@ -28,7 +28,7 @@ Description: Estimates an ecological niche using occurrence data, covariates, an
 License: Apache License (>= 2.0)
 Encoding: UTF-8
 Roxygen: list(markdown = TRUE)
-RoxygenNote: 7.2.1
+RoxygenNote: 7.2.2
 Depends:
     R (>= 3.5.0)
 Imports: 
@@ -43,15 +43,13 @@ Imports:
     grDevices,
     iterators,
     pls,
-    raster,
-    rgeos,
     ROCR,
-    sp,
+    sf,
     sparr,
     spatstat.geom,
-    stats
+    stats,
+    terra
 Suggests: 
-    maptools,
     R.rsp,
     RStoolbox,
     spatstat.data,

diff --git a/NAMESPACE b/NAMESPACE
@@ -34,23 +34,10 @@ importFrom(graphics,plot.new)
 importFrom(graphics,title)
 importFrom(iterators,icount)
 importFrom(pls,cvsegments)
-importFrom(raster,crs)
-importFrom(raster,cut)
-importFrom(raster,extract)
-importFrom(raster,image)
-importFrom(raster,projectRaster)
-importFrom(raster,raster)
-importFrom(raster,rasterToPoints)
-importFrom(raster,reclassify)
-importFrom(raster,values)
-importFrom(rgeos,gBuffer)
-importFrom(sp,CRS)
-importFrom(sp,Polygon)
-importFrom(sp,Polygons)
-importFrom(sp,SpatialPolygons)
-importFrom(sp,bbox)
-importFrom(sp,coordinates)
-importFrom(sp,gridded)
+importFrom(sf,st_bbox)
+importFrom(sf,st_buffer)
+importFrom(sf,st_coordinates)
+importFrom(sf,st_polygon)
 importFrom(sparr,risk)
 importFrom(spatstat.geom,as.solist)
 importFrom(spatstat.geom,im.apply)
@@ -65,3 +52,12 @@ importFrom(spatstat.geom,superimpose)
 importFrom(stats,median)
 importFrom(stats,na.omit)
 importFrom(stats,sd)
+importFrom(terra,classify)
+importFrom(terra,crds)
+importFrom(terra,crs)
+importFrom(terra,extract)
+importFrom(terra,image)
+importFrom(terra,project)
+importFrom(terra,rast)
+importFrom(terra,res)
+importFrom(terra,values)
diff --git a/NEWS.md b/NEWS.md
@@ -1,5 +1,14 @@
 # envi (development version)
 
+# envi v0.1.16.9000
+* Migrated R-spatial dependencies
+* Replaced `raster`, `rgeos`, and `sp` packages in Imports with `terra` and `sf` because of imminent package retirements
+* Removed `maptools` from Suggests (replaced with new internal function `as.im.SpatRaster()`)
+* Thank you, [Roger Bivand](https://github.com/rsbivand), for the notice. Relates to [ndi Issue #3](https://github.com/lance-waller-lab/envi/issues/3)
+* Note: `raster` is a dependency of `RStoolbox` (at present) which is used in the vignette
+* Updated test, examples, vignette, and documentation throughout
+* Added GitHub R-CMD-check
+
 # envi v0.1.15
 * Uwe Ligges suggested (2022-08-26) that some parallel cluster is not cleanly closed in the tests
 * Added `future::plan(future::multisession)` in tests to remove the files in temp directory

diff --git a/R/div_plot.R b/R/div_plot.R
@@ -1,8 +1,8 @@
-#' Prepare an 'im' or 'raster' object for plotting with diverging color palette
+#' Prepare an 'im' or 'SpatRaster' object for plotting with diverging color palette
 #' 
-#' Internal function to convert 'im' object or 'RasterLayer' object to values readable by \code{\link[fields]{image.plot}} function within the \code{\link{plot_obs}}, \code{\link{plot_predict}}, and \code{\link{plot_perturb}} functions. 
+#' Internal function to convert 'im' object or 'SpatRaster' object to values readable by \code{\link[fields]{image.plot}} function within the \code{\link{plot_obs}}, \code{\link{plot_predict}}, and \code{\link{plot_perturb}} functions. 
 #' 
-#' @param input An object of class 'im' or 'RasterLayer' from the \code{\link{lrren}} function.
+#' @param input An object of class 'im' or 'SpatRaster' from the \code{\link{lrren}} function.
 #' @param plot_cols Character string of length three (3) specifying the colors for plotting: 1) presence, 2) neither, and 3) absence from the \code{\link{plot_obs}} function. 
 #' @param midpoint Numeric. The value to center the diverging color palette. 
 #' @param thresh_up Numeric. The upper value to concatenate the color key. The default (NULL) uses the maximum value from \code{input}.
@@ -20,7 +20,7 @@
 #' }
 #' 
 #' @importFrom grDevices colorRampPalette
-#' @importFrom raster raster
+#' @importFrom terra rast
 #'
 #' @keywords internal
 
@@ -33,15 +33,15 @@ div_plot <- function(input,
 
   # Inputs
   if (inherits(input, "im")) {
-    out <- raster::raster(input)
+    out <- terra::rast(input)
   } else { out <- input }
 
   if (length(cols) != 3) {
     stop("The 'cols' argument must be a vector of length 3")
   }
 
-  min_raw_value <- min(out[is.finite(out)], na.rm = TRUE) # minimum absolute value of raster
-  max_raw_value <- max(out[is.finite(out)], na.rm = TRUE) # maximum absolute value of raster
+  min_raw_value <- min(out[is.finite(out)], na.rm = TRUE) # minimum absolute value of SpatRaster
+  max_raw_value <- max(out[is.finite(out)], na.rm = TRUE) # maximum absolute value of SpatRaster
 
   # Restrict spurious log relative risk values
   if (!is.null(thresh_low)) {
@@ -60,8 +60,8 @@ div_plot <- function(input,
   # Identify ramp above and below midpoint
   lowerhalf <- length(out[out < midpoint & !is.na(out)]) # values below 0
   upperhalf <- length(out[out > midpoint & !is.na(out)]) # values above 0
-  min_absolute_value <- min(out[is.finite(out)], na.rm = TRUE) # minimum absolute value of raster
-  max_absolute_value <- max(out[is.finite(out)], na.rm = TRUE) # maximum absolute value of raster
+  min_absolute_value <- min(out[is.finite(out)], na.rm = TRUE) # minimum absolute value of SpatRaster
+  max_absolute_value <- max(out[is.finite(out)], na.rm = TRUE) # maximum absolute value of SpatRaster
 
   # Color ramp parameters
   ## Colors

diff --git a/R/lrren.R b/R/lrren.R
@@ -70,12 +70,11 @@
 #' @importFrom grDevices chull
 #' @importFrom iterators icount
 #' @importFrom pls cvsegments
-#' @importFrom raster extract raster
-#' @importFrom rgeos gBuffer
-#' @importFrom sp bbox coordinates Polygon Polygons SpatialPolygons
+#' @importFrom sf st_bbox st_buffer st_coordinates st_polygon
 #' @importFrom sparr risk
 #' @importFrom spatstat.geom owin ppp
 #' @importFrom stats na.omit
+#' @importFrom terra extract rast values
 #' @export
 #'
 #' @examples
@@ -89,8 +88,8 @@
 #'   grad <- spatstat.data::bei.extra[[2]]
 #'   elev$v <- scale(elev)
 #'   grad$v <- scale(grad)
-#'   elev_raster <- raster::raster(elev)
-#'   grad_raster <- raster::raster(grad)
+#'   elev_raster <- terra::rast(elev)
+#'   grad_raster <- terra::rast(grad)
 #' 
 #' # Presence data
 #'   presence <- spatstat.data::bei
@@ -115,8 +114,10 @@
 #'   obs_locs <- obs_locs[ , c(6, 2, 3, 1, 4, 5)]
 #'   
 #' # Prediction Data
-#'   predict_locs <- data.frame(raster::rasterToPoints(elev_raster))
-#'   predict_locs$layer2 <- raster::extract(grad_raster, predict_locs[, 1:2])
+#'   predict_xy <- terra::crds(elev_raster)
+#'   predict_locs <- as.data.frame(predict_xy)
+#'   predict_locs$elev <- terra::extract(elev_raster, predict_xy)[ , 1]
+#'   predict_locs$grad <- terra::extract(grad_raster, predict_xy)[ , 1]
 #' 
 #' # Run lrren
 #'   test_lrren <- lrren(obs_locs = obs_locs,
@@ -138,7 +139,7 @@ lrren <- function(obs_locs,
                   n_core = 2,
                   poly_buffer = NULL,
                   obs_window = NULL,
-                  verbose = FALSE,
+                  verbose = FALSE, 
                   ...) {
 
   if (verbose == TRUE) { message("Estimating relative risk surfaces\n") }
@@ -148,35 +149,32 @@ lrren <- function(obs_locs,
   # Compute spatial windows
   ## Calculate inner boundary polygon (extent of presence and absence locations in environmental space)
   inner_chull <- concaveman::concaveman(as.matrix(obs_locs[ , 5:6]))
-  inner_chull_pts <- sp::coordinates(inner_chull)
-  inner_chull_pts <- rbind(inner_chull_pts, inner_chull_pts[1, ])
-  inner_chull_poly <- sp::SpatialPolygons(list(sp::Polygons(list(sp::Polygon(inner_chull_pts)), 1)))
-
+  inner_chull_poly <- sf::st_polygon(list(inner_chull))
+
   if (is.null(poly_buffer)) {
-    poly_buffer <- abs(min(diff(sp::bbox(inner_chull_poly)[1, ]), diff(sp::bbox(inner_chull_poly)[2, ])) / 100)
+    poly_buffer <- abs(min(diff(sf::st_bbox(inner_chull_poly)[c(1,3)]), diff(sf::st_bbox(inner_chull_poly)[c(2,4)])) / 100)
   }
 
   # add small buffer around polygon to include boundary points
-  inner_chull_poly_buffer <- rgeos::gBuffer(inner_chull_poly, width = poly_buffer, byid = TRUE)
-  inner_poly <- inner_chull_poly_buffer@polygons[[1]]@Polygons[[1]]@coords
+  inner_chull_poly_buffer <- sf::st_buffer(inner_chull_poly, dist = poly_buffer, byid = TRUE)
+  inner_poly <- sf::st_polygon(list(as.matrix(inner_chull_poly_buffer)))
 
   if (is.null(predict_locs)) {
     outer_chull_poly <- inner_chull_poly_buffer
     outer_poly <- inner_poly
   } else {
     ## Calculate outer boundary polygon (full extent of geographical extent in environmental space)
     if (nrow(predict_locs) > 5000000) { # convex hull
-      outer_chull <- grDevices::chull(x = stats::na.omit(predict_locs)[ , 3], y = stats::na.omit(predict_locs)[ , 4])
-      outer_chull_pts <- predict_locs[c(outer_chull, outer_chull[1]), 3:4]
+      predict_locs_woNAs <- stats::na.omit(predict_locs)
+      outer_chull <- grDevices::chull(x = predict_locs_woNAs[ , 3], y = predict_locs_woNAs[ , 4])
+      outer_chull_pts <- predict_locs_woNAs[c(outer_chull, outer_chull[1]), 3:4]
     } else { # concave hull
-      outer_chull <- concaveman::concaveman(as.matrix(stats::na.omit(predict_locs)[ , 3:4]))
-      outer_chull_pts <- sp::coordinates(outer_chull)
+      outer_chull_pts <- concaveman::concaveman(as.matrix(stats::na.omit(predict_locs[ , 3:4])))
     }
-    outer_chull_pts <- rbind(outer_chull_pts, outer_chull_pts[1, ])
-    outer_chull_poly <- sp::SpatialPolygons(list(sp::Polygons(list(sp::Polygon(outer_chull_pts)), 1)))
-    #add small buffer around polygon to include boundary points
-    outer_chull_poly_buffer <- rgeos::gBuffer(outer_chull_poly, width = poly_buffer, byid = TRUE)
-    outer_poly <- outer_chull_poly_buffer@polygons[[1]]@Polygons[[1]]@coords #extract coordinates of new polygon
+    outer_chull_poly <- sf::st_polygon(list(as.matrix(outer_chull_pts)))
+    # add small buffer around polygon to include boundary points
+    outer_chull_poly_buffer <- sf::st_buffer(outer_chull_poly, dist = poly_buffer, byid = TRUE)
+    outer_poly <- sf::st_polygon(list(as.matrix(outer_chull_poly_buffer)))
   }
 
   if (conserve == FALSE & is.null(predict_locs)) {
@@ -185,8 +183,8 @@ lrren <- function(obs_locs,
   if (conserve == TRUE) { window_poly <- inner_poly } else { window_poly <- outer_poly }
 
   if (is.null(obs_window)) {
-    wind <- spatstat.geom::owin(poly = list(x = rev(window_poly[ , 1]),
-                                            y = rev(window_poly[ , 2])))
+    wind <- spatstat.geom::owin(poly = list(x = rev(sf::st_coordinates(window_poly)[ , 1]),
+                                            y = rev(sf::st_coordinates(window_poly)[ , 2])))
   } else { wind <- obs_window }
 
   # Input Preparation
@@ -207,7 +205,7 @@ lrren <- function(obs_locs,
   obs <- sparr::risk(f = ppp_presence,
                      g = ppp_absence,
                      tolerate = TRUE,
-                     verbose = verbose,
+                     verbose = verbose, 
                      ...)
   bandw <- obs$f$h0
 
@@ -228,17 +226,17 @@ lrren <- function(obs_locs,
       # Project relative risk surface into geographic space
       if (verbose == TRUE) { message("Predicting area of interest") }
 
-    # Convert to semi-continuous raster
-    rr_raster <- raster::raster(obs$rr)
+    # Convert to semi-continuous SpatRaster
+    rr_raster <- terra::rast(obs$rr)
 
-    # Convert to categorical raster
-    pval_raster <- raster::raster(obs$P)
+    # Convert to categorical SpatRaster
+    pval_raster <- terra::rast(obs$P)
 
     # Prediction locations
     extract_points <- cbind(predict_locs[ , 3], predict_locs[ , 4])
     extract_predict <- data.frame("predict_locs" = predict_locs,
-                                  "rr" = raster::extract(rr_raster, extract_points),
-                                  "pval" = raster::extract(pval_raster, extract_points))
+                                  "rr" = terra::extract(rr_raster, extract_points)[ , 1],
+                                  "pval" = terra::extract(pval_raster, extract_points)[ , 1])
 
     output <- list("obs" = obs,
                    "presence" = ppp_presence,
@@ -326,19 +324,19 @@ lrren <- function(obs_locs,
       rand_lrr <- sparr::risk(f = ppp_presence_training,
                               g = ppp_absence_training,
                               tolerate = TRUE,
-                              verbose = FALSE,
+                              verbose = FALSE, 
                               ...)
 
-      ##### Convert to semi-continuous raster
-      rr_raster <- raster::raster(rand_lrr$rr)
-      rr_raster[is.na(rr_raster[])] <- 0 # if NA, assigned null value (log(rr) = 0)
+      ##### Convert to semi-continuous SpatRaster
+      rr_raster <- terra::rast(rand_lrr$rr)
+      rr_raster[is.na(terra::values(rr_raster))] <- 0 # if NA, assigned null value (log(rr) = 0)
 
       ##### Predict testing dataset
       extract_testing <- testing[ , 5:6]
 
       ##### Output for each k-fold
       ###### Record category (semi-continuous) of testing data
-      cv_predictions_rr <- raster::extract(rr_raster, extract_testing)
+      cv_predictions_rr <- terra::extract(rr_raster, extract_testing)[ , 2]
       cv_labels <- testing[ , 4] # Record labels (marks) of testing data
 
       par_results <- list("cv_predictions_rr" = cv_predictions_rr,

diff --git a/R/package.R b/R/package.R
@@ -30,7 +30,7 @@
 #' @aliases envi-package envi 
 #' @docType package
 #' 
-#' @section Dependencies: The 'envi' package relies heavily upon \code{\link{sparr}}, \code{\link{spatstat.geom}}, and \code{\link{raster}}. For a single species (presence/absence data), the spatial relative risk function uses the \code{\link[sparr]{risk}} function. Cross-validation is can be performed in parallel using the \code{\link{future}}, \code{\link{doFuture}}, \code{\link{doRNG}}, and \code{\link{foreach}} packages. Spatial perturbation is performed using the \code{\link[spatstat.geom]{rjitter}} function. Basic visualizations rely on the \code{\link[spatstat.geom]{plot.ppp}} and \code{\link[fields]{image.plot}} functions.
+#' @section Dependencies: The 'envi' package relies heavily upon \code{\link{sparr}}, \code{\link{spatstat.geom}}, \code{\link{sf}}, and \code{\link{terra}}. For a single species (presence/absence data), the spatial relative risk function uses the \code{\link[sparr]{risk}} function. Cross-validation is can be performed in parallel using the \code{\link{future}}, \code{\link{doFuture}}, \code{\link{doRNG}}, and \code{\link{foreach}} packages. Spatial perturbation is performed using the \code{\link[spatstat.geom]{rjitter}} function. Basic visualizations rely on the \code{\link[spatstat.geom]{plot.ppp}} and \code{\link[fields]{image.plot}} functions.
 #' 
 #' @author Ian D. Buller\cr \emph{Environmental Health Sciences, Emory University, Atlanta, Georgia, USA.}\cr
 #' 
@@ -50,10 +50,9 @@ NULL
 #' @importFrom grDevices chull colorRampPalette
 #' @importFrom iterators icount
 #' @importFrom pls cvsegments
-#' @importFrom raster crs cut extract image projectRaster raster rasterToPoints reclassify values
-#' @importFrom rgeos gBuffer
 #' @importFrom ROCR performance prediction
-#' @importFrom sp bbox coordinates CRS gridded Polygon Polygons SpatialPolygons
+#' @importFrom sf st_bbox st_buffer st_coordinates st_polygon
 #' @importFrom spatstat.geom as.solist im.apply marks owin pixellate plot.ppp ppp rjitter setmarks superimpose
 #' @importFrom stats median na.omit sd
+#' @importFrom terra crds crs image project rast res classify values
 NULL