R/spi.R

#'Calculate Standardized Precipitation Index (SPI)
#'
#'Calculate SPI and the drought specifications with the length, the drought 
#'type and the intensity
#'
#'@param  prec_data [zoo] rainfall monthly data in zoo class with date 
#'in \%Y-\%m-\%d
#'@param  time_step [numeric] by default = 12, time step to sum monthly data
#'(1, 3, 6, 9, 12, 24 and 48)
#'@param  distribution [character] distribution of data 
#'(log_Logistic, gamma, grev, genlog, normal)
#'
#'@return list that contains
#'@return \emph{spi} [zoo] zoo with the spi values with date in \%Y-\%m-\%d
#'@return \emph{drought_type} [zoo] zoo with the type of the period for each
#'month
#'@return \emph{drought_number} [data.frame] dataframe with the number of 
#'different period by type
#'\itemize{
#'\item Extwet (spi > 2)\cr
#'\item Verywet (1.99 > spi > 1.5)\cr
#'\item Wet (1.49 > spi > 1)\cr
#'\item Normal (0.99 > spi > -0.99)\cr
#'\item Dry (-1 > spi > -1.49)\cr
#'\item VeryDry (-1.5 > spi > -1.99)\cr
#'\item ExtDry (-2 > spi)
#'}
#'
#'@author Florine Garcia (florine.garcia@gmail.com)
#'@author Pierre L'Hermite (pierrelhermite@yahoo.fr)
#'
#'@examples
#'How to use function
#'
#'@references
#'McKee, T.B., Doesken, N.J., & Kleist, J. (1993)
#'The relationship of drought frequency and duration of time scales.
#'\emph{Eighth Conference on Applied Climatology, American Meteorological Society}
#'\url{http://www.droughtmanagement.info/literature/AMS_Relationship_Drought_Frequency_Duration_Time_Scales_1993.pdf}
#'@seealso
#'\code{\link[piflowtest]{plot_trend}}: plot the index

spi <- function(prec_data, time_step = 12, distribution = "gamma"){
  
  ##__Checking______________________________________________________________####
  # Data input checking
  if (!is.zoo(prec_data)) { stop("prec_data must be a zoo"); return(NULL)}

  # Time step checking
  if (periodicity(prec_data)$scale != "monthly") {
    stop("prec_data must be a monthly serie \n"); return(NULL)
  }
  
  # First month
  firstmonth <- as.numeric(substr(index(prec_data[1]), 6, 7))
  
  # Using SPEI package to calculate spi
  res_spi <- SPEI::spi(coredata(MonthlyData[which(!is.na(MonthlyData))]),
                       scale = time_step, distribution = distribution,
                       na.rm = TRUE)
  spi <- zoo(as.numeric(res_spi$fitted),
              order.by = index(MonthlyData[which(!is.na(MonthlyData))]))
  
  ##__Index analysis________________________________________________________####
  # Drought type and number of drought
  ext_wet <- very_wet <- wet <- normal <- dry <- very_dry <- ext_dry <- 0
  drought_type <- rep(NA, length(spi))
  
  for (i in 1:length(coredata(spi))) {
    if (is.na(coredata(spi)[i])) {
    } else if ((coredata(spi)[i] >= 3)) {
      ext_wet <- ext_wet + 1
      drought_type[i] <- 3
    } else if ((2.99 > coredata(spi)[i]) && (coredata(spi)[i] > 2)) {
      very_wet <- very_wet + 1
      drought_type[i] <- 2
    } else if ((1.99 > coredata(spi)[i]) && (coredata(spi)[i] > 1)) {
      wet <- wet + 1
      drought_type[i] <- 1
    } else if ((0.99 > coredata(spi)[i]) && (coredata(spi)[i] > -0.99)) {
      normal <- normal+1
      drought_type[i] <- 0
    } else if ((-1 >= coredata(spi)[i]) && (coredata(spi)[i] > -1.99)) {
      dry <- dry + 1
      drought_type[i] <- - 1
    } else if ((-2 >= coredata(spi)[i]) && (coredata(spi)[i] > -2.99)) {
      very_dry <- very_dry + 1
      drought_type[i] <- - 2
    } else if ((coredata(spi)[i] <= -3)) {
      ext_dry <- ext_dry + 1
      drought_type[i] <- - 3
    } else {}
  }
  
  drought_number <- rbind.data.frame(ext_wet, very_wet, wet, normal, dry,
                                     very_dry, ext_dry)
  colnames(drought_number) <- c("Rain gauge")
  row.names(drought_number) <- c("Extreme Wet", "Very Wet", "Wet", "Normal",
                                 "Dry", "Very Dry", "Extreme Dry")
  
  # Calculation of the drought length
  length_drought <- numeric()
  
  n <- 0
  p <- 0
  for (ilength in 1:length(spi)) {
    if (is.na(spi[ilength])){
      length_drought[ilength] <- NA
    } else if (spi[ilength] > 0) {
      n <- 0
      p <- p + 1
      length_drought[ilength] <- p
    } else {
      p <- 0
      n <- n - 1 
      length_drought[ilength] <- n
    }
  }
  
  length_zoo <- zoo(as.numeric(length_drought), index(spi))
  
  Resultat <- list(spi = spi, drougth_length = length_zoo,
                   drought_number_type = drought_number, type_time = drought_type)
  return(Resultat)
}
flogrc/piflow documentation built on May 24, 2019, 7:34 a.m.