An error occurred while loading the file. Please try again.
-
Guillaume Perréal authored2c823ad6
# \\\# Copyright 2021-2022 Louis Héraut*1,# Éric Sauquet*2,# Valentin Mansanarez## *1 INRAE, France# louis.heraut@inrae.fr# *2 INRAE, France# eric.sauquet@inrae.fr## This file is part of ash R toolbox.## Ash R toolbox is free software: you can redistribute it and/or# modify it under the terms of the GNU General Public License as# published by the Free Software Foundation, either version 3 of the# License, or (at your option) any later version.## Ash R toolbox is distributed in the hope that it will be useful, but# WITHOUT ANY WARRANTY; without even the implied warranty of# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU# General Public License for more details.## You should have received a copy of the GNU General Public License# along with ash R toolbox.# If not, see <https://www.gnu.org/licenses/>.# ///### Rcode/plotting/datasheet.R## Regroups all the graphical tools to generates the datasheets. More precisely, the 'datasheet_panel' function manages all the call for each station of the different graphical functions that generates info header, time serie visualisation and trend analysis graphs for every variable. It also deals with the arranging of all the plots in a single PDF page.# Sourcing R filesource(file.path('Rcode', 'processing', 'analyse.R'), encoding='UTF-8') # hydrographsource(file.path('Rcode', 'plotting', 'shortcut.R'), encoding='UTF-8')## 1. DATASHEET PANEL MANAGER ________________________________________# Manages datasheets creations for all stations. Makes the call to# the different headers, trend analysis graphs and realises arranging# every plots.datasheet_panel = function (list_df2plot, df_meta, trend_period, mean_period, linetype_per, axis_xlim, colorForce, info_header, time_header, foot_note, layout_matrix, info_height, time_ratio, var_ratio, foot_height, paper_size, resources_path, df_shapefile, logo_dir, PRlogo_file, AEAGlogo_file, INRAElogo_file, FRlogo_file, outdirTmp_pdf, outdirTmp_png, df_page=NULL) { # The percentage of augmentation and diminution of the min # and max limits for y axis lim_pct = 10 # Number of variable/plot nbVar = length(list_df2plot) # Get all different stations code Code = levels(factor(df_meta$code)) nCode = length(Code) if (!is.null(trend_period)) { # Convert 'trend_period' to list trend_period = as.list(trend_period) # Number of trend period nPeriod_trend = length(trend_period)7172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140
# Extracts the min and the max of the mean trend for all the station
res = short_trendExtremes(list_df2plot, Code, nPeriod_trend, nbVar, nCode, colorForce)
minTrendValue = res$min
maxTrendValue = res$max
}
# Blank vector to store the max number of digit of label for
# each station
NspaceMax = c()
# For all the station
for (code in Code) {
# Default max digit
NspaceMax_code = 0
# If the time header is given it adds one to the number of plot
nbVarMod = nbVar + as.numeric(!is.null(time_header))
# For all type of graph
for (i in 1:nbVarMod) {
if (i > nbVar) {
# Extracts the data serie corresponding to the code
df_data_code = time_header[time_header$code == code,]
type = 'sévérité'
} else {
# Extracts the data corresponding to the current variable
df_data = list_df2plot[[i]]$data
# Extracts the type corresponding to the current variable
type = list_df2plot[[i]]$type
# Extracts the data corresponding to the code
df_data_code = df_data[df_data$code == code,]
}
# If variable type is date
if (type == 'saisonnalité') {
# The number of digit is 6 because months are display
# with 3 characters
Nspace = 6
# If it is a flow variable
} else if (type == 'sévérité' | type == 'data' | type == 'pluviométrie' | type == 'température' | type == 'évapotranspiration') {
# Gets the max number of digit on the label
maxtmp = max(df_data_code$Value, na.rm=TRUE)
# Taking into account of the augmentation of
# max for the window
maxtmp = maxtmp * (1 + lim_pct/100)
# If the max is greater than 10
if (maxtmp >= 10) {
# The number of digit is the magnitude plus
# the first number times 2
Nspace = (get_power(maxtmp) + 1)*2
# Plus spaces between thousands hence every 8 digits
Nspace = Nspace + as.integer(Nspace/8)
# If the max is less than 10 and greater than 1
} else if (maxtmp < 10 & maxtmp >= 1) {
# The number of digit is the magnitude plus
# the first number times 2 plus 1 for the dot
# and 2 for the first decimal
Nspace = (get_power(maxtmp) + 1)*2 + 3
# If the max is less than 1 (and obviously more than 0)
} else if (maxtmp < 1) {
# Fixes the number of significant decimals to 3
maxtmp = signif(maxtmp, 3)
# The number of digit is the number of character
# of the max times 2 minus 1 for the dots that
# count just 1 space
Nspace = nchar(as.character(maxtmp))*2 - 1
}
}
141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210
# If it is the temporary max number
if (Nspace > NspaceMax_code) {
# Stores it
NspaceMax_code = Nspace
}
}
# Stores the max digit number for labels of a station
NspaceMax = c(NspaceMax, NspaceMax_code)
}
# For all the station
for (k in 1:nCode) {
# Gets the code
code = Code[k]
# Print code of the station for the current plotting
print(paste("Datasheet for station : ", code,
" (", round(k/nCode*100, 0), " %)",
sep=''))
# Number of header (is info and time serie are needed)
nbh = as.numeric(!is.null(info_header)) + as.numeric(!is.null(time_header))
nbp = max(layout_matrix, na.rm=TRUE)
# Actualises the number of plot
nbg = nbp + nbh + as.numeric(foot_note)
# Opens a blank list to store plot
P = vector(mode='list', length=nbg)
for (id in 1:nbg) {
P[[id]] = void
}
# If the info header is needed
if (!is.null(info_header)) {
if ("data.frame" %in% class(info_header)) {
# Extracts the data serie corresponding to the code
info_header_code = info_header[info_header$code == code,]
to_do = 'all'
} else {
info_header_code = NULL
to_do = info_header
}
# Gets the info plot
Hinfo = info_panel(list_df2plot,
df_meta,
trend_period=trend_period,
mean_period=mean_period,
periodHyd=mean_period[[1]],
df_shapefile=df_shapefile,
codeLight=code,
df_data_code=info_header_code,
to_do=to_do)
# Stores it
P[[1]] = Hinfo
}
# If the time header is given
if (!is.null(time_header)) {
# Extracts the data serie corresponding to the code
time_header_code = time_header[time_header$code == code,]
if (is.null(axis_xlim)) {
# Gets the limits of the time serie
axis_xlim_code = c(min(time_header_code$Date),
max(time_header_code$Date))
} else {
axis_xlim_code = axis_xlim
211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280
}
# Gets the time serie plot
Htime = time_panel(time_header_code, df_trend_code=NULL,
trend_period=trend_period,
axis_xlim=axis_xlim_code, missRect=TRUE,
unit2day=365.25, var='Q', type='sévérité',
grid=TRUE, ymin_lim=0,
NspaceMax=NspaceMax[k],
first=TRUE, lim_pct=lim_pct)
# Stores it
P[[2]] = Htime
} else {
axis_xlim_code = axis_xlim
}
# Computes the number of column of plot asked on the datasheet
nbcol = ncol(as.matrix(layout_matrix))
# For all variable
for (i in 1:nbVar) {
# Extracts the data corresponding to the current variable
df_data = list_df2plot[[i]]$data
# Extracts the trend corresponding to the
# current variable
df_trend = list_df2plot[[i]]$trend
unit2day = list_df2plot[[i]]$unit2day
missRect = list_df2plot[[i]]$missRect
# Extract the variable of the plot
var = list_df2plot[[i]]$var
type = list_df2plot[[i]]$type
# Extracts the data corresponding to the code
df_data_code = df_data[df_data$code == code,]
# Extracts the trend corresponding to the code
df_trend_code = df_trend[df_trend$code == code,]
# Blank vector to store color
color = c()
if (!is.null(trend_period)) {
# For all the period
for (j in 1:nPeriod_trend) {
# If the trend is significant
# if (df_trend_code$p[j] <= alpha | colorForce){
if (df_trend_code$p[j] <= alpha){
# Extract start and end of trend periods
Start = df_trend_code$period_start[j]
End = df_trend_code$period_end[j]
# Extracts the corresponding data for the period
df_data_code_per =
df_data_code[df_data_code$Date >= Start
& df_data_code$Date <= End,]
# Same for trend
df_trend_code_per =
df_trend_code[df_trend_code$period_start == Start
& df_trend_code$period_end == End,]
# Computes the number of trend analysis selected
Ntrend = nrow(df_trend_code_per)
# If there is more than one trend on the same period
if (Ntrend > 1) {
# Takes only the first because they are similar
df_trend_code_per = df_trend_code_per[1,]
}
# If it is a flow variable
if (type == 'sévérité') {
# Computes the mean of the data on the period
281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350
dataMean = mean(df_data_code_per$Value,
na.rm=TRUE)
# Normalises the trend value by the mean
# of the data
trendValue = df_trend_code_per$trend / dataMean
# If it is a date variable
} else if (type == 'saisonnalité' | type == 'pluviométrie' | type == 'température' | type == 'évapotranspiration') {
trendValue = df_trend_code_per$trend
}
# gets the color corresponding to the mean trend
color_res = get_color(trendValue,
minTrendValue[j, i],
maxTrendValue[j, i],
palette_name='perso',
reverse=TRUE)
# Stores it temporarily
colortmp = color_res
# Otherwise
} else {
# Stores the default grey color
colortmp = 'grey85'
}
# Stores the color
color = append(color, colortmp)
grid = FALSE
}
}
if (var != 'sqrt(Q)' & var != 'Q') {
grid = FALSE
ymin_lim = NULL
} else {
grid = TRUE
ymin_lim = 0
}
if (is.null(time_header) & i == 1) {
first = TRUE
} else {
first = FALSE
}
# Computes the time panel associated to the current variable
p = time_panel(df_data_code, df_trend_code, var=var,
type=type, linetype_per=linetype_per,
alpha=alpha, colorForce=colorForce,
missRect=missRect, trend_period=trend_period,
mean_period=mean_period,
axis_xlim=axis_xlim_code,
unit2day=unit2day, grid=grid,
ymin_lim=ymin_lim, color=color,
NspaceMax=NspaceMax[k], first=first,
last=(i == nbVar),
lim_pct=lim_pct)
# Stores the plot
P[[i+nbh]] = p
}
if (!is.null(df_page)) {
section = 'Fiche station'
subsection = code
n_page = df_page$n[nrow(df_page)] + 1
df_page = bind_rows(
df_page,
tibble(section=section,
subsection=subsection,
351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420
n=n_page))
}
if (foot_note) {
footName = 'fiche station'
if (is.null(df_page)) {
n_page = k
}
foot = foot_panel(footName, n_page, resources_path,
logo_dir, PRlogo_file,
AEAGlogo_file, INRAElogo_file,
FRlogo_file, foot_height)
P[[nbg]] = foot
}
# Convert the 'layout_matrix' to a matrix if it is not already
layout_matrix = as.matrix(layout_matrix)
# Number of element of the matrix
nel = nrow(layout_matrix)*ncol(layout_matrix)
# Gets the place where there is NA value
idNA = which(is.na(layout_matrix), arr.ind=TRUE)
LM = layout_matrix
# Adds non existing plot is where there is NA
LM[idNA] = seq(max(layout_matrix, na.rm=TRUE) + 1,
max(layout_matrix, na.rm=TRUE) + 1 +
nel)
# Shifts all plots to be coherent with the adding of header
LM = LM + nbh
if (!is.null(time_header)) {
id_time = nbh
LM = rbind(nbh, LM)
} else {
id_time = NA
time_ratio = 0
}
if (!is.null(info_header)) {
id_info = nbh - as.numeric(!is.null(time_header))
LM = rbind(nbh - as.numeric(!is.null(time_header)), LM)
} else {
id_info = NA
info_ratio = 0
}
if (foot_note) {
id_foot = length(LM) + 1
LM = rbind(LM, id_foot)
} else {
id_foot = max(LM) + 1
foot_height = 0
}
# If paper format is A4
if (paper_size == 'A4') {
width = 21
height = 29.7
} else if (is.vector(paper_size) & length(paper_size) > 1) {
width = paper_size[1]
height = paper_size[2]
}
LMcol = ncol(LM)
LMrow = nrow(LM)
LM = rbind(rep(99, times=LMcol), LM, rep(99, times=LMcol))
LMrow = nrow(LM)
LM = cbind(rep(99, times=LMrow), LM, rep(99, times=LMrow))
421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490
LMcol = ncol(LM)
margin_size = 0.5
Norm_ratio = height * (time_ratio + var_ratio*nbp) / (height - 2*margin_size - foot_height - info_height)
time_height = height * time_ratio / Norm_ratio
var_height = height * var_ratio / Norm_ratio
Hcut = LM[, 2]
heightLM = rep(0, times=LMrow)
heightLM[Hcut == id_info] = info_height
heightLM[Hcut == id_time] = time_height
heightLM[Hcut > nbh & Hcut < id_foot] = var_height
heightLM[Hcut == id_foot] = foot_height
heightLM[Hcut == 99] = margin_size
col_width = (width - 2*margin_size) / (LMcol - 2)
Wcut = LM[(nrow(LM)-1),]
widthLM = rep(col_width, times=LMcol)
widthLM[Wcut == 99] = margin_size
# Plot the graph as the layout
plot = grid.arrange(grobs=P, layout_matrix=LM,
heights=heightLM, widths=widthLM)
# Saving
ggsave(plot=plot,
path=outdirTmp_pdf,
filename=paste(as.character(code), '.pdf', sep=''),
width=width, height=height, units='cm', dpi=100)
# Saving
ggsave(plot=plot,
path=outdirTmp_png,
filename=paste(as.character(code), '.png', sep=''),
width=width, height=height, units='cm', dpi=400)
}
return (df_page)
}
## 2. PANEL FOR THE DATASHEET __________________________________
### 2.1. Time panel __________________________________________________
time_panel = function (df_data_code, df_trend_code, var, type,
linetype_per='solid', alpha=0.1,
colorForce=FALSE, missRect=FALSE,
unit2day=365.25, trend_period=NULL,
mean_period=NULL, axis_xlim=NULL, grid=TRUE,
ymin_lim=NULL, color=NULL, NspaceMax=NULL,
first=FALSE, last=FALSE, lim_pct=10) {
# Compute max and min of flow
maxQ = max(df_data_code$Value, na.rm=TRUE)
minQ = min(df_data_code$Value, na.rm=TRUE)
spread = maxQ - minQ
nTick = 6
maxQ_win = maxQ + spread*lim_pct/100
minQ_win = minQ - spread*lim_pct/100
if (minQ_win < 0) {
minQtmp_lim = 0
} else {
minQtmp_lim = minQ_win
}
491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560
if (!is.null(ymin_lim)) {
minQ_win = ymin_lim
}
spreadtmp = maxQ_win - minQtmp_lim
breakQtmp = spreadtmp / (nTick - 1)
GradQ_10 = c(0, 1, 1.5, 2, 2.5, 3, 4, 5, 10)
Grad = GradQ_10 * 10^get_power(breakQtmp)
dist = abs(Grad - breakQtmp)
idGrad = which.min(dist)
breakQ = Grad[idGrad]
if (is.null(ymin_lim)) {
Grad = GradQ_10 * 10^get_power(minQtmp_lim)
Grad[Grad > minQtmp_lim] = NA
dist = abs(Grad - minQtmp_lim)
idGrad = which.min(dist)
minQ_lim = Grad[idGrad]
} else {
minQ_lim = ymin_lim
}
maxQ_list = c()
i = 1
maxQtmp = minQ_lim
while (maxQtmp <= maxQ_win) {
maxQtmp = minQ_lim + i*breakQ
i = i + 1
}
maxQ_lim = maxQtmp
# If x axis limits are specified
if (!is.null(axis_xlim)) {
axis_xlim = as.Date(axis_xlim)
minor_minDatetmp_lim = axis_xlim[1]
minor_maxDatetmp_lim = axis_xlim[2]
# Otherwise
} else {
minor_minDatetmp_lim = as.Date(df_data_code$Date[1])
minor_maxDatetmp_lim =
as.Date(df_data_code$Date[length(df_data_code$Date)])
}
minor_minDatetmp_lim = as.numeric(format(minor_minDatetmp_lim, "%Y"))
minor_maxDatetmp_lim = as.numeric(format(minor_maxDatetmp_lim, "%Y"))
minDatetmp_lim = minor_minDatetmp_lim
maxDatetmp_lim = minor_maxDatetmp_lim
nTick = 8
spreadtmp = minor_maxDatetmp_lim - minor_minDatetmp_lim
breakDatetmp = spreadtmp / (nTick - 1)
GradDate_10 = c(1, 2.5, 5, 10)
Grad = GradDate_10 * 10^get_power(breakDatetmp)
dist = abs(Grad - breakDatetmp)
idGrad = which.min(dist)
breakDate = Grad[idGrad]
listDate = seq(round(minDatetmp_lim, -1)-10^(get_power(breakDate)+1),
round(maxDatetmp_lim, -1)+10^(get_power(breakDate)+1),
by=breakDate)
minDate_lim = listDate[which.min(abs(listDate - minDatetmp_lim))]
561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630
maxDate_lim = listDate[which.min(abs(listDate - maxDatetmp_lim))]
minDate_lim = as.Date(paste(minDate_lim, '-01-01', sep=''))
maxDate_lim = as.Date(paste(maxDate_lim, '-01-01', sep=''))
minor_breakDatetmp = breakDate / 5
GradMinorDate_10 = c(1, 2, 5, 10)
Grad = GradMinorDate_10 * 10^get_power(minor_breakDatetmp)
dist = abs(Grad - minor_breakDatetmp)
idGrad = which.min(dist)
minor_breakDate = Grad[idGrad]
listDate = seq(round(minor_minDatetmp_lim,
-1) - 10^(get_power(minor_breakDate)+1),
round(minor_maxDatetmp_lim,
-1) + 10^(get_power(minor_breakDate)+1),
by=minor_breakDate)
minor_minDate_lim =
listDate[which.min(abs(listDate - minor_minDatetmp_lim))]
minor_maxDate_lim =
listDate[which.min(abs(listDate - minor_maxDatetmp_lim))]
minor_minDate_lim = as.Date(paste(round(minor_minDate_lim),
'-01-01', sep=''))
minor_maxDate_lim = as.Date(paste(round(minor_maxDate_lim),
'-01-01', sep=''))
# Open new plot
p = ggplot() + theme_ash
# Margins
if (first) {
p = p +
theme(plot.margin=margin(t=2.5, r=0, b=3, l=0, unit="mm"))
} else if (last) {
p = p +
theme(plot.margin=margin(t=2, r=0, b=0, l=0, unit="mm"))
} else if (first & last) {
p = p +
theme(plot.margin=margin(t=2.5, r=0, b=0, l=0, unit="mm"))
} else {
p = p +
theme(plot.margin=margin(t=2, r=0, b=2, l=0, unit="mm"))
}
## Sub period background ##
if (!is.null(trend_period)) {
# Convert trend period to list if it is not
trend_period = as.list(trend_period)
# Fix a disproportionate minimum for period
Imin = 10^99
# For all the sub period of analysis in 'trend_period'
for (per in trend_period) {
# Compute time interval of period
I = interval(per[1], per[2])
# If it is the smallest interval
if (I < Imin) {
# Store it
Imin = I
# Fix min period of analysis
trend_period_min = as.Date(per)
}
}
minPer = trend_period_min[1]
maxPer = trend_period_min[2]
631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700
# If there is an 'axis_lim'
if (!is.null(axis_xlim)) {
# If the temporary start of period is smaller
# than the fix start of x axis limit
if (minPer < axis_xlim[1]) {
# Set the start of the period to the start of
# the x axis limit
minPer = axis_xlim[1]
}
# If the temporary end of period plus one year
# is smaller than the fix end of x axis limit
if (maxPer + years(1) < axis_xlim[2]) {
# Add one year the the temporary end of period
maxPer = maxPer + years(1)
} else {
# Set the start of the period to the start of
# the x axis limit
maxPer = axis_xlim[2]
}
# If there is no 'axis_lim'
} else {
if (minPer < min(df_data_code$Date)) {
minPer = min(df_data_code$Date)
}
if (maxPer > max(df_data_code$Date)) {
maxPer = max(df_data_code$Date)
}
}
# Draw rectangle to delimiting the sub period
p = p +
geom_rect(aes(xmin=minPer,
ymin=minQ_win,
xmax=maxPer,
ymax= maxQ_win),
linetype=0, fill='grey97')
}
## Mean step ##
# If there is a 'mean_period'
if (!is.null(mean_period)) {
# Convert 'mean_period' to list
mean_period = as.list(mean_period)
# Number of mean period
nPeriod_mean = length(mean_period)
# Blank tibble to store variable in order to plot
# rectangle for mean period
plot_mean = tibble()
# Blank tibble to store variable in order to plot
# upper limit of rectangle for mean period
plot_line = tibble()
# For all mean period
for (j in 1:nPeriod_mean) {
# Get the current start and end of the sub period
xmin = as.Date(mean_period[[j]][1])
xmax = as.Date(mean_period[[j]][2])
# Extract the data corresponding to this sub period
df_data_code_per =
df_data_code[df_data_code$Date >= xmin
& df_data_code$Date <= xmax,]
# If the min over the sub period is greater
# than the min of the entier period and
# it is not the first sub period
if (xmin > min(df_data_code$Date) & j != 1) {
# Substract 6 months to be in the middle of
701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770
# the previous year
xmin = add_months(xmin, -6)
}
# If it is not a flow or sqrt of flow time serie and
# it is the first period
if (var != 'sqrt(Q)' & var != 'Q' & j == 1) {
# If there is an x axis limit
if (!is.null(axis_xlim)) {
# If the min of the period is before the x axis min
if (xmin < axis_xlim[1]) {
# The min for the sub period is the x axis
xmin = axis_xlim[1]
}
}
}
# If the max over the sub period is smaller
# than the max of the entier period and
# it is not the last sub period
if (xmax < max(df_data_code$Date) & j != nPeriod_mean) {
# Add 6 months to be in the middle of
# the following year
xmax = add_months(xmax, 6)
}
# If it is not a flow or sqrt of flow time serie and
# it is the last period
if (var != 'sqrt(Q)' & var != 'Q' & j == nPeriod_mean) {
# If there is an x axis limit
if (!is.null(axis_xlim)) {
# If the max of the period plus 1 year
# is smaller thant the max of the x axis limit
if (xmax + years(1) < axis_xlim[2]) {
# Add one year to the max to include
# the entire last year graphically
xmax = xmax + years(1)
} else {
# The max of this sub period is the max
# of the x axis limit
xmax = axis_xlim[2]
}
}
# If there is no axis limit
# } else {
# # Add one year to the max to include
# # the entire last year graphically
# xmax = xmax + years(1)
# }
}
# Mean of the flow over the sub period
ymax = mean(df_data_code_per$Value, na.rm=TRUE)
# Create temporary tibble with variable
# to create rectangle for mean step
plot_meantmp = tibble(xmin=xmin, xmax=xmax,
ymin=minQ_win, ymax=ymax, period=j)
# Bind it to the main tibble to store it with other period
plot_mean = bind_rows(plot_mean, plot_meantmp)
# Create vector for the upper limit of the rectangle
abs = c(xmin, xmax)
ord = c(ymax, ymax)
# Create temporary tibble with variable
# to create upper limit for rectangle
plot_linetmp = tibble(abs=abs, ord=ord, period=j)
# Bind it to the main tibble to store it with other period
plot_line = bind_rows(plot_line, plot_linetmp)
}
771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840
# Plot rectangles
p = p +
geom_rect(data=plot_mean,
aes(xmin=xmin, ymin=ymin,
xmax=xmax, ymax=ymax),
linetype=0, fill='grey93')
# Plot upper line for rectangle
p = p +
geom_line(data=plot_line,
aes(x=abs, y=ord, group=period),
color='grey85',
size=0.15)
# for all the sub periods except the last one
for (i in 1:(nPeriod_mean - 1)) {
# Computes the time difference in days between periods
dPeriod = abs(as.Date(mean_period[[i+1]][1]) - as.Date(mean_period[[i]][2]))
if (dPeriod < 10) {
# The x limit is the x max of the ith rectangle
xLim = plot_mean$xmax[i]
# The y limit of rectangle is the max of
# the two neighboring mean step rectangle
yLim = max(c(plot_mean$ymax[i], plot_mean$ymax[i+1]))
# Make a tibble to store data
plot_lim = tibble(x=c(xLim, xLim), y=c(minQ_win, yLim))
# Plot the limit of rectangles
p = p +
geom_line(data=plot_lim, aes(x=x, y=y),
linetype='dashed', size=0.15,
color='grey85')
} else {
# Takes the x and y limits for the ith rectangle
xLim_i = plot_mean$xmax[i]
yLim_i = plot_mean$ymax[i]
# Takes the x and y limits for the i+1th rectangle
xLim_i1 = plot_mean$xmin[i+1]
yLim_i1 = plot_mean$ymax[i+1]
# Make a tibble to store data
plot_lim = tibble(x_i=c(xLim_i, xLim_i),
y_i=c(minQ_win, yLim_i),
x_i1=c(xLim_i1, xLim_i1),
y_i1=c(minQ_win, yLim_i1))
# Plot the limit of rectangles
p = p +
geom_line(data=plot_lim, aes(x=x_i, y=y_i),
linetype='dashed', size=0.15,
color='grey85') +
geom_line(data=plot_lim, aes(x=x_i1, y=y_i1),
linetype='dashed', size=0.15,
color='grey85')
}
}
}
### Grid ###
if (grid) {
# If there is no axis limit
if (is.null(axis_xlim)) {
# The min and the max is set by
# the min and the max of the date data
xmin = min(df_data_code$Date)
xmax = max(df_data_code$Date)
} else {
# Min and max is set with the limit axis parameter
xmin = axis_xlim[1]
xmax = axis_xlim[2]
841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910
}
# Create a vector for all the y grid position
ygrid = seq(minQ_win, maxQ_win, breakQ)
# Blank vector to store position
ord = c()
abs = c()
# For all the grid element
for (i in 1:length(ygrid)) {
# Store grid position
ord = c(ord, rep(ygrid[i], times=2))
abs = c(abs, xmin, xmax)
}
# Create a tibble to store all the position
plot_grid = tibble(abs=as.Date(abs), ord=ord)
# Plot the y grid
p = p +
geom_line(data=plot_grid,
aes(x=abs, y=ord, group=ord),
color='grey85',
size=0.15)
}
### Data ###
# If it is a square root flow or flow
if (var == 'sqrt(Q)' | var == 'Q') {
# Plot the data as line
p = p +
geom_line(aes(x=df_data_code$Date, y=df_data_code$Value),
color='grey20',
size=0.3,
lineend="round")
} else {
# Plot the data as point
p = p +
geom_point(aes(x=df_data_code$Date, y=df_data_code$Value),
shape=19, color='grey50', alpha=1,
stroke=0, size=1)
}
### Missing data ###
# If the option is TRUE
if (missRect) {
# Remove NA data
NAdate = df_data_code$Date[is.na(df_data_code$Value)]
# Get the difference between each point of date data without NA
dNAdate = diff(NAdate)
# If difference of day is not 1 then
# it is TRUE for the beginning of each missing data period
NAdate_Down = NAdate[append(Inf, dNAdate) != 1]
# If difference of day is not 1 then
# it is TRUE for the ending of each missing data period
NAdate_Up = NAdate[append(dNAdate, Inf) != 1]
# Plot the missing data period
p = p +
geom_rect(aes(xmin=NAdate_Down,
ymin=minQ_win,
xmax=NAdate_Up,
ymax=maxQ_win),
linetype=0, fill='Wheat', alpha=0.4)
}
### Trend ###
# If there is trends
if (!is.null(df_trend_code)) {
# Extract start and end of trend periods
Start = df_trend_code$period_start
End = df_trend_code$period_end
# Get the name of the different period
911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980
UStart = levels(factor(Start))
UEnd = levels(factor(End))
# Compute the max of different start and end
# so the number of different period
nPeriod_trend = max(length(UStart), length(UEnd))
# Blank tibble to store trend data and legend data
plot_trend = tibble()
leg_trend = tibble()
# For all the different period
for (i in 1:nPeriod_trend) {
# Extracts the corresponding data for the period
df_data_code_per =
df_data_code[df_data_code$Date >= Start[i]
& df_data_code$Date <= End[i],]
# Computes the mean of the data on the period
dataMean = mean(df_data_code_per$Value,
na.rm=TRUE)
# Get the trend associated to the first period
df_trend_code_per =
df_trend_code[df_trend_code$period_start == Start[i]
& df_trend_code$period_end == End[i],]
# Number of trend selected
Ntrend = nrow(df_trend_code_per)
# If the number of trend is greater than a unique one
if (Ntrend > 1) {
# Extract only the first hence it is the same period
df_trend_code_per = df_trend_code_per[1,]
}
# Search for the index of the closest existing date
# to the start of the trend period of analysis
iStart = which.min(abs(df_data_code$Date - Start[i]))
# Same for the end
iEnd = which.min(abs(df_data_code$Date - End[i]))
# Get the start and end date associated
xmin = df_data_code$Date[iStart]
xmax = df_data_code$Date[iEnd]
# If there is a x axis limit
if (!is.null(axis_xlim)) {
# If the min of the current period
# is smaller than the min of the x axis limit
if (xmin < axis_xlim[1]) {
# The min of the period is the min
# of the x axis limit
xmin = axis_xlim[1]
}
# Same for end
if (xmax > axis_xlim[2]) {
xmax = axis_xlim[2]
}
}
# Create vector to store x data
abs = c(xmin, xmax)
# Convert the number of day to the unit of the period
abs_num = as.numeric(abs) / unit2day
# Compute the y of the trend
ord = abs_num * df_trend_code_per$trend +
df_trend_code_per$intercept
# Create temporary tibble with variable to plot trend
# for each period
981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050
plot_trendtmp = tibble(abs=abs, ord=ord, period=i)
# Bind it to the main tibble to store it with other period
plot_trend = bind_rows(plot_trend, plot_trendtmp)
# If there is a x axis limit
if (!is.null(axis_xlim)) {
# The x axis limit is selected
codeDate = axis_xlim
} else {
# The entire date data is selected
codeDate = df_data_code$Date
}
# The y limit is stored in a vector
codeValue = c(minQ_win, maxQ_win)
# Position of the x beginning and end of the legend symbol
x = gpct(1.5, codeDate, shift=TRUE)
xend = x + gpct(3, codeDate)
# Spacing between legend symbols
dy = gpct(9, codeValue, min_lim=ymin_lim)
# Position of the y beginning and end of the legend symbol
y = gpct(92, codeValue,
min_lim=ymin_lim, shift=TRUE) - (i-1)*dy
yend = y
# Position of x for the beginning of the associated text
xt = xend + gpct(1, codeDate)
# Position of the background rectangle of the legend
xminR = x - gpct(1, codeDate)
yminR = y - gpct(5, codeValue, min_lim=ymin_lim)
# If it is a flow variable
if (type == 'sévérité') {
xmaxR = x + gpct(32.5, codeDate)
# If it is a date variable
} else if (type == 'saisonnalité' | type == 'pluviométrie' | type == 'température' | type == 'évapotranspiration') {
xmaxR = x + gpct(20.5, codeDate)
}
ymaxR = y + gpct(5, codeValue, min_lim=ymin_lim)
# Gets the trend
trend = df_trend_code_per$trend
# Gets the p value
pVal = df_trend_code_per$p
if (pVal <= alpha) {
colorLine = color[i]
colorLabel = color[i]
} else {
colorLine = 'grey85'
colorLabel = 'grey85'
}
# Computes the mean trend
trendMean = trend/dataMean
# Computes the magnitude of the trend
power = get_power(trend)
# Converts it to character
powerC = as.character(power)
# If the power is positive
if (powerC >= 0) {
# Adds a space in order to compensate for the minus
# sign that sometimes is present for the other periods
spaceC = ' '
# Otherwise
} else {
# No space is added
spaceC = ''
}
1051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120
# Gets the power of ten of magnitude
brk = 10^power
# Converts trend to character for sientific expression
trendC = as.character(format(round(trend / brk, 2),
nsmall=2))
# If the trend is positive
if (trendC >= 0) {
# Adds two space in order to compensate for the minus
# sign that sometimes is present for the other periods
trendC = paste(' ', trendC, sep='')
}
# Converts mean trend to character
trendMeanC = as.character(format(round(trendMean*100, 2),
nsmall=2))
if (trendMeanC >= 0) {
# Adds two space in order to compensate for the minus
# sign that sometimes is present for the other periods
trendMeanC = paste(' ', trendMeanC, sep='')
}
# Create temporary tibble with variable to plot legend
leg_trendtmp = tibble(x=x, xend=xend,
y=y, yend=yend,
xt=xt,
colorLine=colorLine,
colorLabel=colorLabel,
trendC=trendC,
powerC=powerC,
spaceC=spaceC,
trendMeanC=trendMeanC,
xminR=xminR, yminR=yminR,
xmaxR=xmaxR, ymaxR=ymaxR,
period=i)
# Bind it to the main tibble to store it with other period
leg_trend = bind_rows(leg_trend, leg_trendtmp)
}
if (length(linetype_per) < nPeriod_trend) {
linetype_per = rep(linetype_per, times=nPeriod_trend)
}
linetypeLeg_per = linetype_per
linetypeLeg_per[linetype_per == 'longdash'] = '33'
linetypeLeg_per[linetype_per == 'dashed'] = '22'
linetypeLeg_per[linetype_per == 'dotted'] = '11'
# For all periods
for (i in 1:nPeriod_trend) {
# Extract the trend of the current sub period
leg_trend_per = leg_trend[leg_trend$period == i,]
# Plot the background for legend
p = p +
geom_rect(data=leg_trend_per,
aes(xmin=xminR,
ymin=yminR,
xmax=xmaxR,
ymax=ymaxR),
linetype=0, fill='white', alpha=0.3)
# Get the character variable for naming the trend
colorLine = leg_trend_per$colorLine
colorLabel = leg_trend_per$colorLabel
trendC = leg_trend_per$trendC
powerC = leg_trend_per$powerC
spaceC = leg_trend_per$spaceC
trendMeanC = leg_trend_per$trendMeanC
# If it is a flow variable
1121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190
if (type == 'sévérité') {
# Create the name of the trend
label = bquote(bold(.(trendC)~'x'~'10'^{.(powerC)}*.(spaceC))~'['*m^{3}*'.'*s^{-1}*'.'*an^{-1}*']'~~bold(.(trendMeanC))~'[%.'*an^{-1}*']')
# If it is a date variable
} else if (type == 'saisonnalité') {
# Create the name of the trend
label = bquote(bold(.(trendC)~'x'~'10'^{.(powerC)}*.(spaceC))~'[jour.'*an^{-1}*']')
} else if (type == 'pluviométrie' | type == 'évapotranspiration') {
# Create the name of the trend
label = bquote(bold(.(trendC)~'x'~'10'^{.(powerC)}*.(spaceC))~'[mm.'*an^{-1}*']')
} else if (type == 'température') {
# Create the name of the trend
label = bquote(bold(.(trendC)~'x'~'10'^{.(powerC)}*.(spaceC))~'[°C.'*an^{-1}*']')
}
# Plot the trend symbole and value of the legend
p = p +
annotate("segment",
x=leg_trend_per$x, xend=leg_trend_per$xend,
y=leg_trend_per$y, yend=leg_trend_per$yend,
color=colorLine,
linetype=linetypeLeg_per[i],
lwd=0.8,
lineend="round") +
annotate("text",
label=label, size=2.8,
x=leg_trend_per$xt, y=leg_trend_per$y,
hjust=0, vjust=0.5,
color=colorLabel)
}
# For all periods
for (i in 1:nPeriod_trend) {
# Extract the trend of the current sub period
plot_trend_per = plot_trend[plot_trend$period == i,]
# Plot the line of white background of each trend
p = p +
geom_line(data=plot_trend_per,
aes(x=abs, y=ord),
color='white',
linetype='solid',
size=1.5,
lineend="round")
}
# For all periods
for (i in 1:nPeriod_trend) {
# Extract the trend of the current sub period
plot_trend_per = plot_trend[plot_trend$period == i,]
# Plot the line of trend
p = p +
geom_line(data=plot_trend_per,
aes(x=abs, y=ord),
color=color[i],
linetype=linetype_per[i],
size=0.75,
lineend="round")
}
}
# Y axis title
# If it is a flow variable
if (type == 'sévérité' | var == 'Q') {
p = p +
ylab(bquote(bold(.(var))~~'['*m^{3}*'.'*s^{-1}*']'))
} else if (var == 'sqrt(Q)') {