R/ancillary_data_functions.r

In khufkens/ceipr: Centre on Emission Inventories and Projections processing code

#' Add eurostat population numbers check
#'
#' @param df CEIP data frame generated by ceip_totals()
#' @return warning / stop message upon failure
#' @keywords emission, voc, population, data checks
#' @export
check_for_totals <- function(df) {
  # check if the data class is correct
  if(!any(class(df) == "ceipr_totals")){
    if(any(class(df) == "ceipt_data")) {
      warning("Data is of class ceipr_data, adding meta_data to this (raw) data is not recommended. Use ceip_totals to convert to totals")
    } else {
      stop("Data is not of class ceipr_totals, not valid ceipr data!")
    }
  }
}

#' Add eurostat population numbers to CEIP data frame.
#'
#' Warning: only available from 2007!
#'
#' @param df CEIP data frame generated by ceip_read() (of class ceipr_data)
#' @return Merges the two databases
#' @keywords emission, voc, population
#' @export

ceip_add_population <- function(df) {
  # sanity check:
  if(!all(c("iso2","year") %in% names(df))) {
    stop("Data must contain an iso2 (country code) and year variable.")
  }

  suppressMessages(
    # Load the data set for EU population data
    EU_population_data <- eurostat::get_eurostat("tps00001") %>%
      # convert the timestamp to a year, for better referencing
      dplyr::mutate(year=strtoi(format(time,'%Y')),geo=as.character(geo)) %>%
      # rename some variables to make things easier
      dplyr::rename(population = values,eurostat=geo) %>%
      # add country names and EU status
      dplyr::left_join(countrycode::codelist_panel,by=c("year","eurostat")) %>%
      # narrow the results down
      dplyr::select(iso2 = iso2c,country=country.name.en,population,eu28,year)
  )
  return(
    dplyr::left_join(df, EU_population_data, by = c("iso2","year"))
  )
}

#' Calculate per capita emissions
#'
#' @param df CEIP data frame generated by ceip_read() (of class ceipr_data)
#' and supplemented with eurostat data using ceip_add_population()
#' @return Merges the two databases
#' @keywords emission, voc, population
#' @export

ceip_population_emissions <- function(df) {
  # sanity check:
  if(!all(c("total_emissions","population") %in% names(df))) {
    stop("Data does not contain total_emissions and/or population. Did you run ceip_totals and ceip_add_population on your data?")
  }

  # calculate the emissions per capita
  df$emissions_pp <- with(df, 10^6 * total_emissions/population)
  # add unit
  df$unit_pp <- "g"
  return(df)
}

#' Summarize emission totals
#'
#' @param df CEIP data frame generated by ceip_read() (of class ceipr_data)
#' @return a dataframe with emissions per country
#' @export

ceip_totals <- function(df) {
  if(!all(c("iso2","year","pollutant","sector_abbr","unit") %in% names(df))) {
    stop("Data does not appear to be ceipr_data.")
  }

  totals <- df %>%
      dplyr::group_by(iso2,year,pollutant,sector,sector_abbr,unit) %>%
      dplyr::summarise(total_emissions = sum(emission))
  class(totals) <- c(class(totals),"ceipr_totals")
  return(totals)
}