R/flip-layers.R

In atlantistools: Process and Visualise Output from Atlantis Models

#' Flip layers for visualization.
#'
#' Within Atlantis the water column id 0 is the water column closest to the sediment.
#' In order to simplify graphical interpretation of vertical plots this order
#' is reversed. The surface layer is 1 by default. The sediment layer id is equal
#' to the number of total layers. Please note that this is only used for graphical
#' display.
#'
#' @param data dataframe with columns polygon and layer. layer id is based on
#' atlantis output (0 = layer closest to the sediment)
#' @return dataframe with flipped layerids. 1 = surface.
#' @export
#'
#' @examples
#' data <- rbind(expand.grid(species = "sp1", polygon = 0, layer = 0:7),
#'               expand.grid(species = "sp1", polygon = 1, layer = 0:4),
#'               expand.grid(species = "sp1", polygon = 2, layer = 0:2),
#'               expand.grid(species = "sp1", polygon = 3, layer = c(0:3, 7)))
#' data$atoutput <- runif(nrow(data), min = 0, max = 2)
#' flip_layers(data)

flip_layers <- function(data) {
  if (any(!c("polygon", "layer") %in% names(data))) stop("Columns polygon and layer not present in data.")

  # Create a df with fliped layers and normal layers based on the input dataframe.
  df <- unique(dplyr::select_(data, .dots = c("polygon", "layer")))
  sediment <- max(df$layer)
  df_fliped <- split(df, df$polygon) %>%
    lapply(add_fliped_layers, sed = sediment) %>%
    dplyr::bind_rows()

  # Combine original dataframe with fliped layers and replace columns.
  data_fliped <- data %>%
    dplyr::left_join(df_fliped, by = c("polygon", "layer"))
  data_fliped$layer <- NULL
  names(data_fliped)[names(data_fliped) == "layer_fliped"] <- "layer"

    return(data_fliped)
}

add_fliped_layers <- function(df, sed) {
  df <- dplyr::arrange_(df, ~layer)
  if (any(df$layer == sed)) {
    layer_fliped <- c(rev(df$layer[df$layer != sed]), sed) + 1
  } else {
    layer_fliped <- rev(df$layer) + 1
  }
  df$layer_fliped <- layer_fliped
  return(df)
}