R/raster.resid.R

In danlwarren/ENMTools: Analysis of Niche Evolution using Niche and Distribution Models

#' raster.resid Measure standardized residuals from a linear regression between two rasters.
#'
#' This function builds a linear model for the relationship between two rasters, and returns the linear model
#' and a spatial raster of the residuals from that model.
#'
#' @param x A raster or RasterLayer object, or ENMTools model object containing a suitability raster.
#' @param y Another raster  or RasterLayer object, or ENMTools model object containing a suitability raster.
#'
#' @return results A list containing a raster of residuals from a linear regression between the two supplied rasters and the linear model itself.
#'
#' @keywords correlation raster residuals
#'
#' @examples
#' data(euro.worldclim)
#' raster.resid(euro.worldclim[[1]], euro.worldclim[[2]])


raster.resid <- function(x, y){

  if(inherits(x, "enmtools.model")){
    x <- x$suitability
  }

  if(inherits(y, "enmtools.model")){
    y <- y$suitability
  }

  resid.raster <- y

  x <- terra::values(x)
  y <- terra::values(y)
  this.lm <- lm(x ~ y)
  resids <- this.lm$residuals/this.lm$fitted.values

  resid.raster[!is.na(resid.raster)] <- resids

  return(list(residuals = resid.raster, lm = this.lm))
}