The comprehensive Precipitation Generator

ComprehensivePrecipitationGenerator(
  station = c("T0001", "T0010", "T0099"),
  prec_all,
  mean_climate_prec = NULL,
  year_max = 1990,
  year_min = 1961,
  leap = TRUE,
  nmonth = 12,
  cpf = NULL,
  verbose = TRUE,
  p = 1,
  type = "none",
  lag.max = NULL,
  ic = "AIC",
  activateVARselect = FALSE,
  exogen = NULL,
  exogen_sim = NULL,
  is_exogen_gaussian = FALSE,
  year_max_sim = year_max,
  year_min_sim = year_min,
  mean_climate_prec_sim = NULL,
  onlygeneration = FALSE,
  varmodel = NULL,
  type_quantile = 3,
  qnull = NULL,
  valmin = 0.5,
  step = 0,
  n_GPCA_iteration = 0,
  n_GPCA_iteration_residuals = n_GPCA_iteration,
  sample = NULL,
  extremes = TRUE,
  exogen_all = NULL,
  exogen_all_col = station,
  no_spline = FALSE,
  nscenario = 1,
  seed = NULL,
  noise = NULL
)

Arguments

station: character vector of the IDs of the considered meteorological stations
prec_all: data frame containing daily precipitation of all meteorological stations. See PRECIPITATION defined in the trentino dataset for formatting.
mean_climate_prec: a matrix containing monthly mean daily precipitation for the considered station. If it is NULL, it is calculated. See input of is.monthly.climate
year_max: start year of the recorded (calibration) period
year_min: end year of the recorded (calibration) period
leap: logical variables. If it is TRUE (default)(recommended), leap years are considered, otherwise all years have 365 days
nmonth: number of months in one year (default is 12)
cpf: see normalizeGaussian_severalstations
verbose: logical variable
p, type, lag.max, ic, activateVARselect: see respective input parameter on getVARmodel
exogen: data frame or matrix containing the (normalized or not) exogenous variables (predictors) for the recorded (calibration) period.
exogen_sim: data frame or matrix containing the (normalized or not) exogenous variables (predictors) for the simulation period. Default is NULL. If it is NULL, it is replaced with exogen within the function.
is_exogen_gaussian: logical value. If TRUE, exogen_sim and exogen are given as already normalized variables, otherwhise they are not normalized. Default is FALSE
year_max_sim: last year of the simulation period. Default is equal to year_max
year_min_sim: first year of the simulation period. Default is equal to year_min
mean_climate_prec_sim: a matrix containing monthly mean daily precipitation for the simulation period. If is NULL (Default), it is set equal to mean_climate_prec.
onlygeneration: logical value. If TRUE the VAR model varmodel is given as input and only random generation is done, otherwise (default) is calculated from measured data
varmodel: the comprehensinve VAR model as a varest2 S4 object or a NULL object. If NULL (default), the comprehensinve VAR is estimated from measured data within the function, otherwise it is given as input and only random generation is done.
type_quantile: see type on quantile
step: see normalizeGaussian_severalstations. Default is 0.
n_GPCA_iteration: number of iterations of Gaussianization process for data. Default is 0 (no Gaussianization)
n_GPCA_iteration_residuals: number of iterations of Gaussianization process for VAR residuals. Default is 0 (no Gaussianization)
sample, extremes, qnull, valmin: see normalizeGaussian_severalstations
exogen_all: data frame containing exogenous variable formatted like prec_all. Default is NULL. It is alternative to exogen and if it not NULL,is_exogen_gaussian is automatically set FALSE
exogen_all_col: vector of considered columns of exogen_all. Default is station.
no_spline: logical value. See splineInterpolateMonthlytoDailyforSeveralYears. Default is TRUE.
nscenario: number of generated scenarios for daily maximum and minimum temperature
seed: seed for stochastic random generation see set.seed.
noise: stochastic noise to add for variabile generation. Default is NULL. See newVARmultieventRealization. Not used in case that nscenario>1.

Value

A list of the following variables:

prec_mes matrix containing measured daily precipitation (the data is copied by the measured data given as input for the period and the station considered for varmodel estimation)

prec_spline matrix containing climatic "spline-interpolated" daily preciptation from mean_climate_prec

data_prec matrix containing normalized measured precipitation variable

prec_gen matrix containing generated daily precipitation [mm]

prec_spline_sim matrix containing climatic "spline-interpolated" daily preciptation from mean_climate_prec_sim

data_prec_gen matrix containing normalized generated precipitation variable

mean_climate_prec matrix containing monthly means of daily precipitation (historical scenario)

mean_climate_prec_sim matrix containing monthly means of daily precipitation (predicted/simulated scenario)

var a varest object containing the used VAR model

Note

It pre-processes and generates a multi-site precipitation fields. It uses getVARmodel. Detailed examples can be viewed of this function in this presentation. Unfortunately, using this approach, the spatial correlations are underestimated. This is due to the persinstence of zeros in the precipitation records. This problem is known in literature and can be solved in the future versions of RMAWGEN. See the R code for further details

Author

Emanuele Cordano, Emanuele Eccel

Examples