R/bru.spatial.R

In inlabru-org/inlabru: Bayesian Latent Gaussian Modelling using INLA and Extensions

# Convert data frame to SpatialLinesDataFrame
#
#
# @aliases sfill
# @export
# @param data A SpatialGridDataFrame or SpatialPixelDataFrame
# @param where Spatial*DataFrame of locations for which to fill in values from \code{data}. If NULL, use \code{data} to determine the locations.
# @return Spatial object

sfill <- function(data, where = NULL) {
  check_spatstat("spatstat.geom")

  if (is.null(where)) {
    where <- data
  }
  vallist <- list()
  for (k in seq_len(ncol(data@data))) {
    dpoints <- SpatialPoints(data)
    vals <- data@data[, k]
    dpoints <- dpoints[!is.na(vals), ]
    vals <- vals[!is.na(vals)]

    data.ow <- spatstat.geom::owin(
      range(sp::coordinates(dpoints)[, 1]),
      range(sp::coordinates(dpoints)[, 2])
    )
    data.ppp <- spatstat.geom::as.ppp(sp::coordinates(dpoints), data.ow)
    where.ow <- spatstat.geom::owin(
      range(sp::coordinates(where)[, 1]),
      range(sp::coordinates(where)[, 2])
    )
    where.ppp <- spatstat.geom::as.ppp(sp::coordinates(where), where.ow)

    nn <- spatstat.geom::nncross(where.ppp, data.ppp)[, "which"]
    vallist[[k]] <- vals[nn]
  }
  ret <- data.frame(do.call(data.frame, vallist))
  colnames(ret) <- colnames(data@data)
  ret <- sp::SpatialPixelsDataFrame(where, data = ret)
}


#' Convert data frame to SpatialLinesDataFrame
#'
#' A line in 2D space is defined by a start and an and point, each associated with 2D coordinates.
#' This function takes a `data.frame` as input and assumes that each row defines a line in space.
#' In order to do so, the data frame must have at least four columns and the `start.cols` and
#' `end.cols` parameters must be used to point out the names of the columns that define
#' the start and end coordinates of the line. The data is then converted to a
#' `SpatialLinesDataFrame` `DF`. If a coordinate reference system `crs` is provided
#' it is attached to `DF`. If also `to.crs` is provided, the coordinate system of `DF`
#' is transfromed accordingly. Additional columns of the input data, e.g. covariates,
#' are retained and attached to `DF`.
#'
#'
#' @aliases sline
#' @export
#' @param data A data.frame
#' @param start.cols Character array poitning out the columns of `data` that hold the start points of the lines
#' @param end.cols Character array poitning out the columns of `data` that hold the end points of the lines
#' @param crs Coordinate reference system of the original `data`
#' @param to.crs Coordinate reference system for the SpatialLines ouput.
#' @return SpatialLinesDataFrame
#'
#' @examples
#' \donttest{
#' # Create a data frame defining three lines
#' lns <- data.frame(
#'   xs = c(1, 2, 3), ys = c(1, 1, 1), # start points
#'   xe = c(2, 3, 4), ye = c(2, 2, 2)
#' ) # end points
#'
#'
#' # Conversion to SpatialLinesDataFrame without CRS
#' spl <- sline(lns,
#'   start.cols = c("xs", "ys"),
#'   end.cols = c("xe", "ye")
#' )
#'
#' if (require(ggplot2, quietly = TRUE)) {
#'   # Plot the lines
#'   ggplot() +
#'     gg(spl)
#' }
#' }
#'
sline <- function(data, start.cols, end.cols, crs = CRS(as.character(NA)), to.crs = NULL) {
  sp <- as.data.frame(data[, start.cols])
  ep <- as.data.frame(data[, end.cols])

  colnames(sp) <- c("x", "y")
  colnames(ep) <- c("x", "y")

  lilist <- lapply(seq_len(nrow(sp)), function(k) {
    Lines(list(Line(rbind(sp[k, ], ep[k, ]))), ID = k)
  })
  spl <- SpatialLines(lilist, proj4string = crs)

  df <- data[, setdiff(names(data), c(start.cols, end.cols))]
  rownames(df) <- seq_len(nrow(df))

  slines <- SpatialLinesDataFrame(spl, data = df)

  # If requested, change CRS
  if (!is.null(to.crs)) slines <- fm_transform(slines, to.crs)

  slines
}


#' Convert a data.frame of boundary points into a SpatialPolgonsDataFrame
#'
#' A polygon can be described as a sequence of points defining the polygon's boundary.
#' When given such a sequence (anti clockwise!) this function creates a
#' SpatialPolygonsDataFrame holding the polygon decribed. By default, the
#' first two columns of `data` are assumed to define the x and y coordinates
#' of the points. This behavior can ba changed using the `cols` parameter, which
#' points out the names of the columns holding the coordinates. The coordinate
#' reference system of the resulting spatial polygon can be set via the `crs`
#' paraemter. Posterior conversion to a different CRS is supported using the
#' `to.crs` parameter.
#'
#' @aliases spoly
#' @export
#' @param data A data.frame of points describing the boundary of the polygon
#' @param cols Column names of the x and y coordinates within the data
#' @param crs Coordinate reference system of the points
#' @param to.crs Coordinate reference system for the SpatialLines ouput.
#' @return SpatialPolygonsDataFrame
#'
#' @examples
#' \donttest{
#' # Create data frame of boundary points (anti clockwise!)
#' pts <- data.frame(
#'   x = c(1, 2, 1.7, 1.3),
#'   y = c(1, 1, 2, 2)
#' )
#'
#' # Convert to SpatialPolygonsDataFrame
#' pol <- spoly(pts)
#'
#' if (require(ggplot2, quietly = TRUE)) {
#'   # Plot it!
#'   ggplot() +
#'     gg(pol)
#' }
#' }
#'
spoly <- function(data, cols = colnames(data)[1:2], crs = fm_CRS(), to.crs = NULL) {
  po <- sp::Polygon(data[, cols], hole = FALSE)
  pos <- sp::Polygons(list(po), ID = "tmp")
  predpoly <- sp::SpatialPolygons(list(pos), proj4string = crs)
  df <- data.frame(weight = 1)
  rownames(df) <- "tmp"
  spoly <- sp::SpatialPolygonsDataFrame(predpoly, data = df)

  # If requested, change CRS
  if (!is.null(to.crs)) spoly <- fm_transform(spoly, to.crs)
  spoly
}