R/data_cleaning.R

In GregoireButruille/FishNmix2: Assess fish abiotic compatibility

#' Title Remove suspicious occurence points
#'
#' @param data A data frame downloaded from GBIF
#' @param minlat Minimum latitude
#' @param maxlat Maximum latitude
#' @param minlong Minimum longitude
#' @param maxlong Maximum longitude
#' @param check.out Check outliers or no (take a long time for a large number of occurences)
#'
#'@importFrom dplyr select
#'@importFrom dplyr filter
#'@import countrycode
#'@import CoordinateCleaner
#'@import rnaturalearthdata
#' @return
#' @export
#'
#' @examples
data_cleaning <- function(data, minlat, maxlat, minlong, maxlong, check.out = TRUE){

  #keep only the columns of interest and set spatial extent
  data_cl <- data%>%
    dplyr::select(species, decimalLongitude, decimalLatitude, gbifID, countryCode)%>%
    filter(decimalLatitude<maxlat)%>%
    filter(decimalLatitude>(minlat))%>%
    filter(decimalLongitude<maxlong)%>%
    filter(decimalLongitude>(minlong))

  #prepare data for cleaning
  names(data_cl)[2:3] <- c("decimallongitude", "decimallatitude") #these are default names for latitude and longitude in the following functions
  data_cl$countryCode <-  countrycode(data_cl$countryCode, origin =  'iso2c', destination = 'iso3c') #iso 2 --> iso 3 changes countrycode from 2 letters to 3 letters (ex : FR --> FRA) to be able to use cc_count()

  #country code puts Na for countrycodes not matched unambiguously (XK and ZZ = Kosovo and undefined countries), remove the Na
  data_cl <- na.omit(data_cl)

  #removes suspicious points (bufffer = range in meters)
  data_cl <- data_cl%>%
    cc_val()%>%  #invalid values
    cc_cap(buffer=10000)%>%   #capitals
    cc_cen(buffer = 1000)%>%  #country centroids
    cc_coun(iso3 = "countryCode")%>%  #country mismatches
    cc_gbif(buffer=1000)%>%  #gbif HeadQuarters
    cc_inst(buffer=100)%>%  #institutions
    cc_sea()  #sea

  if (check.out == TRUE){
    #check outliers
    data_cl <- data_cl%>%   #/!\ takes quite a long time /!\
      cc_outl(
        species = "species", #check species by species
        method = "distance",
        tdi = 1000,          #if a point is more than 1000km away from every other points from the same species, it is removed
        value = "clean",
        thinning = TRUE,
        thinning_res = 0.5
      )
  }

  #Rename latitude and longitude
  names(data_cl)[2:3] <- c("decimalLongitude", "decimalLatitude")

  return(data_cl)
}