amber (version 1.0.3)

correlationMatrix: Correlation matrix for statistical metrics computed by AMBER

Description

This function produces a correlation matrices for mean values, \(bias\), \(rmse\), \(crmse\), \(phase\), and corresponding scores. The input data consist of netCDF files produced by scores.grid.time and scores.grid.notime.

Usage

correlationMatrix(metric, fileList = FALSE, myFiles = NA,
  myRows = NA, myColumns = NA, inputDir, outputDir = FALSE,
  significanceLevel = 0.01, plot.width = 8, plot.height = 6.8,
  plot.margin = c(10, 10, 1, 4))

Arguments

metric

A string that indicates for what statistical metric the correlation matrix should be computed. Options are 'mod-mean', ref-mean', bias', 'crmse', 'phase', 'bias-score', 'crmse-score', 'phase-score', or 'iav-score'.

fileList

Logical, FALSE by default. Should be set to TRUE if user defines a list of files that shall be included in the analysis. If set to FALSE, the function will search for all files whose names include the string of the metric defined above.

myFiles

Names of NetCDF files that shall be included in the analysis. Only relevant if fileList is TRUE.

myRows

Optional: the user can highlight relations between variables by specifying variable names along the rows and columns of a matrix. Those relations are then highlighted by plotting a polygon around the corresponding grid cells and by plotting the corresponding value.

myColumns

Optional: Same as myRows but for variable names listed along the columns of the matrix.

inputDir

A string that gives the location of NetCDF files produced by scores.grid.time, e.g. '/home/project/study'.

outputDir

A string that gives the output directory, e.g. '/home/project/study'. The output will only be written if the user specifies an output directory.

significanceLevel

A number that gives the desired significance level of a correlation, e.g. 0.01

plot.width

A number that gives the plot width, e.g. 8

plot.height

A number that gives the plot height, e.g. 6.8

plot.margin

An R object that gives the plot margin, e.g. c(10, 10, 1, 4)

Value

A list with the Spearman correlation coefficient and corresponding p-values, and a Figure of the correlation matrix

Examples

# NOT RUN {
library(amber)
library(classInt)
library(doParallel)
library(foreach)
library(Hmisc)
library(latex2exp)
library(ncdf4)
library(parallel)
library(raster)
library(rgdal)
library(rgeos)
library(scico)
library(sp)
library(stats)
library(utils)
library(viridis)
library(xtable)

inputDir <- paste(system.file('extdata', package = 'amber'), 'zonalMeanStats', sep = '/')
correlationMatrix(metric = 'bias', inputDir = inputDir)

# Alternatively, use a list of file names
myFiles <- c('GPP-FluxCom-bias.nc', 'LAI-AVHRR-bias.nc', 'ALBS-CERES-bias.nc')

# You can specify certain relationships to highlight them in your correlation matrix
myRows <- c('LAI.AVHRR', 'ALBS.CERES')
myColumns <- c('GPP.FluxCom', 'GPP.FluxCom')

correlationMatrix(metric = 'bias', fileList = TRUE, myFiles = myFiles,
myRows = myRows, myColumns = myColumns, inputDir = inputDir)
# }
# NOT RUN {
 #donttest
# }