In ecor/RGENERATE: Tools to Generate Vector Time Series

knitr::opts_chunk$set(echo = TRUE)

Rationale

The rationale of this vignette is the presentation of a stochastic generation of daily minimum and maximum temperature in two or more sites makes use of RMAWGEN and RGENERATE packages. The illustrated procedure refers to trentino dataset (RMAWGEN package), ad after loading the necessary packages and calling the trentino dataset (RMAWGEN package), is organized as follows:

Selection of a generally 30-year long period within the dataset;
Preparation of input data for the model setup, e.g. detection of temperature anomalies;
Calibration of the stochastic generations, i.e. creation of a GPCAvarest2 object instance for the model whose parameters like order of autocorrelation and number of GPCA iterations are inserted as function arguments;
Usage of the model to generate a new time series of temperature anomalies.

This tutorial aims at generating a time series of temperature anomalies stating from a time series of observations. The observations were cleaned from interpolations of the averages of each month of the year representing the actual climate conditions. The generated time series maintains the same statistical properties of the one of the observed anomalies. The anomalies can be added to different scenarios of climate conditions.

Necessary Package

rm(list=ls())

library(RMAWGEN)
library(RGENERATE)
library(magrittr)

Dataset

The dateset used is trentino dataset from RMAWGEN package, it contains new data frames TEMPERATURE_MAX and TEMPERATURE_MIN :

data(trentino)
str(TEMPERATURE_MAX)
str(TEMPERATURE_MIN)

Reference Period

year_min <- 1978
year_max <- 2007
origin <- sprintf("%04d-1-1",year_min) 
origin_input <- origin
origin_output <- origin

####

station <- c("T0090","T0083")

Settings

sample <- "monthly"

###
n_GPCA_iteration <- 15 
n_GPCA_iteration_residuals <- 15
p <- 2

###
param <- setComprehensiveTemperatureGeneratorParameters(station=station,Tx_all=TEMPERATURE_MAX,Tn_all=TEMPERATURE_MIN,year_min=year_min,year_max=year_max,sample=sample)
std_tm=param[['stdTm']] ## Standard deviation for "mean" 
param$data_original

###

Stochastic Model

exogen <- NULL ## GAUSSIANIZATION!!!


###
model <- getVARmodel(param$data_for_var,suffix=c("_T1","_T2"),sep="",p=p,exogen=exogen,n_GPCA_iteration_residuals=n_GPCA_iteration_residuals,n_GPCA_iteration=n_GPCA_iteration) 
###

Generation of Anomalies

gen1 <- generate(model,n=nrow(param$data_for_var),names=names(param$data_for_var)) 

gen1 <- gen1 %>% normalizeGaussian_severalstations(data=param$data_original,inverse=TRUE,type=3,sample=sample,origin_x=origin_output,origin_data=origin_input)
##res_multigen <- normalizeGaussian_severalstations(x=res_multigen0,data=original_data,inverse=TRUE,type=type,sample=sample,origin_x=origin_x,origin_data=origin_data,extremes=extremes)
###

Observed and Generated Anomalies

###
for (i in 1:ncol(gen1)) {

  print(ks.test(param$data_original[,i],gen1[,i]))
}
##getVARmodel <-
##  function (data,suffix=c("_Tx","_Tn"),sep="",p=1,type="none",season=NULL,exogen=NULL,lag.max=NULL,ic="AIC",activateVARselect=FALSE,na.rm=TRUE,
##            n_GPCA_iteration=0,n_GPCA_iteration_residuals=n_GPCA_iteration,extremes=TRUE) { 

#####
#####