aboutsummaryrefslogtreecommitdiff
path: root/man/FOCUS_Step_12_scenarios.Rd
blob: f58dea79596d5894fe85d2ab41b4dda5d31d6fcf (plain) (blame)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
% Generated by roxygen2: do not edit by hand
% Please edit documentation in R/FOCUS_Step_12_scenarios.R
\docType{data}
\name{FOCUS_Step_12_scenarios}
\alias{FOCUS_Step_12_scenarios}
\title{Step 1/2 scenario data as distributed with the FOCUS Step 1/2 calculator}
\format{A list containing the scenario names in a character vector called 'names',
  the drift percentiles in a matrix called 'drift', interception percentages in
  a matrix called 'interception' and the runoff/drainage percentages for Step 2
  calculations in a matrix called 'rd'.}
\description{
The data were extracted from the scenario.txt file using the R code shown below.
The text file is not included in the package as its licence is not clear.
}
\examples{

\dontrun{
  # This is the code that was used to extract the data
  scenario_path <- "inst/extdata/FOCUS_Step_12_scenarios.txt"
  scenarios <- readLines(scenario_path)[9:38]
  FOCUS_Step_12_scenarios <- list()
  sce <- read.table(text = scenarios, sep = "\\t", header = TRUE, check.names = FALSE,
    stringsAsFactors = FALSE)
  FOCUS_Step_12_scenarios$names = sce$Crop
  rownames(sce) <- sce$Crop
  FOCUS_Step_12_scenarios$drift = sce[, 3:11]
  FOCUS_Step_12_scenarios$interception = sce[, 12:15]
  sce_2 <- readLines(scenario_path)[41:46]
  rd <- read.table(text = sce_2, sep = "\\t")[1:2]
  rd_mat <- matrix(rd$V2, nrow = 3, byrow = FALSE)
  dimnames(rd_mat) = list(Time = c("Oct-Feb", "Mar-May", "Jun-Sep"),
                          Region = c("North", "South"))
  FOCUS_Step_12_scenarios$rd = rd_mat
  save(FOCUS_Step_12_scenarios, file = "data/FOCUS_Step_12_scenarios.RData")
}

# And this is the resulting data
FOCUS_Step_12_scenarios
}
\keyword{datasets}

Contact - Imprint