In NaiaraLopezRojo/globalIRmap: What the Package Does (One Line, Title Case)

#Possible palettes #1: #538797, #C2D9CD, #FF842A https://www.instagram.com/p/B9En6h2gbJ5/
#Possible palette #2: #238189, #3FA8B3, #E8F3EE, #FF674D https://www.instagram.com/p/BxZZu82gBm3/
#File structure
rootdir <- rprojroot::find_root(rprojroot::has_dir("src"))
srcdir <- file.path(rootdir, 'src', 'globalIRmap')
datdir <- file.path(rootdir, 'data')
resdir <- file.path(rootdir, 'results')

#Source packages functions
library(drake)
library(flextable)
library(grid)
source(file.path(srcdir, 'R/IRmapping_packages.R'))
source(file.path(srcdir, 'R/IRmapping_functions.R'))
source(file.path(srcdir, 'R/IRmapping_plan.R'))

loadd(gpredsdt)
loadd(rivpred)
loadd(tasks_u10)
loadd(tasks_featsel_u10)
loadd(tasks_featsel_o1)
loadd(IRpop)

Geographical distribution of reference streamgauging stations

Statistics for paper

Number of gauges in total: \
r gpredsdt[, .N] \
Number of perennial gauges used in analysis (average number of years of data): \ r gpredsdt[intermittent_o1800==0, .N](r gpredsdt[intermittent_o1800==0, mean(totalYears_kept_o1800)]) \
Number of intermittent gauges taking in account only post-1800, mDur >= 1 (average number of years of data): \ r gpredsdt[intermittent_o1800==1, .N](r gpredsdt[intermittent_o1800==1, mean(totalYears_kept_o1800)]) \
Number of intermittent gauges taking in account only post-1800, mDur > 0 (average number of years of data): \ r gpredsdt[mDur_o1800>0, .N] \
Number of intermittent gauges taking in account only post-1800, mDur >= 30 (average number of years of data): \ r gpredsdt[mDur_o1800>=30, .N] \
Number of reaches in network: \ r rivpred[dis_m3_pyr > 0.1 & INLAKEPERC < 1, .N] \
Average length of reaches in network: \ r rivpred[dis_m3_pyr > 0.1 & INLAKEPERC < 1, mean(LENGTH_KM*(1-INLAKEPERC))] \
Total length of reaches in network: \ r rivpred[dis_m3_pyr > 0.1 & INLAKEPERC < 1, sum(LENGTH_KM*(1-INLAKEPERC))] \
Total number of predictor variables: \ r length(tasks_u10$classif$feature_names) \
Total number of predictor variables kept in model for gauges < 10 m3/s: \ r length(tasks_featsel_u10[[2]]$feature_names) \
Total number of predictor variables kept in model for gauges >= 1 m3/s: \ r length(tasks_featsel_o1[[2]]$feature_names) \
Number of people living nearest to an IRES: \ r flextable(IRpop) \

remove(gpredsdt)
remove(rivpred)
remove(tasks_u10)
remove(tasks_featsel_u10)
remove(tasks_featsel_o1)
remove(IRpop)
gc()

Threshold sensitivity

loadd(threshold_sensitivity)

png(paste0(file.path(resdir, 'figures//thresholdsensitivty_'), 
           format(Sys.Date(), '%Y%m%d'), '.png'), 
    width=6, height=6, units='in', res=600)
print(threshold_sensitivity$gperf)
dev.off()

print(threshold_sensitivity$predbounds)

IRES extrapolation

loadd(IRESextra)
loadd(netlength_extra)
print("Number of climate-basins sub-units occurrences: ")
print(IRESextra$preds[,.N])


print(paste0(
    'For mdur >= 1, statistically extrapolating the prevalence of intermittence in rivers < 0.1 m3/s, we predict that ',
    round(100*IRESextra$preds[
      , sum(percinter_all_GAM*cumL_pred, na.rm=T)/sum(cumL_pred, na.rm=T)]),
    '% of rivers >= 0.01 m3/s are intermittent'
  ))

print(paste0('For a total river length of ',IRESextra$preds[, sum(cumL_pred)], ' km'
             ))


print(paste0(
  'For mdur >= 1, assuming that rivers < 0.1 m3/s are as intermittent as those [0.1, 0.2),  we predict that ',
  round(100*IRESextra$preds[
    , sum(percinter_all_conservative*cumL_pred, na.rm=T)/sum(cumL_pred, na.rm=T)]),
  '% of rivers >= 0.01 m3/s are intermittent'
))


print('Example plots')
png(paste0(file.path(resdir, 'figures//extraIRESplot_62_18_'), 
           format(Sys.Date(), '%Y%m%d'), '.png'), 
    width=5, height=5, units='in', res=600)
print(IRESextra$plot_62_18)
dev.off()
png(paste0(file.path(resdir, 'figures//extraIRESplot_14_17_'), 
           format(Sys.Date(), '%Y%m%d'), '.png'), 
    width=5, height=5, units='in', res=600)
print(IRESextra$plot_14_17)
dev.off()

png(paste0(file.path(resdir, 'figures//extraLENGTHplot_62_18_'), 
           format(Sys.Date(), '%Y%m%d'), '.png'), 
    width=5, height=5, units='in', res=600)
print(netlength_extra$plot_62_18)
dev.off()
png(paste0(file.path(resdir, 'figures//extraLENGTHplot_14_17_'), 
           format(Sys.Date(), '%Y%m%d'), '.png'), 
    width=5, height=5, units='in', res=600)
print(netlength_extra$plot_14_17)
dev.off()



loadd(IRESextra_mdur30)
print(paste0(
    'For mdur >= 30, statistically extrapolating the prevalence of intermittence in rivers < 0.1 m3/s, we predict that ',
    round(100*IRESextra_mdur30$preds[
      , sum(percinter_all_GAM*cumL_pred, na.rm=T)/sum(cumL_pred, na.rm=T)]),
    '% of rivers >= 0.01 m3/s are intermittent'
  ))

remove(IRESextra)
remove(IRESextra_mdur30)
gc()

Gauges map

loadd(gauges_plot)

png(paste0(file.path(resdir, 'figures//gauges_plot'), 
           format(Sys.Date(), '%Y%m%d'), '.png'), 
    width=178, height=178, units='mm', res=1000)
print(gauges_plot)
dev.off()


remove(gauges_plot)
gc()

WaterGAP stats

loadd(watergap_stats)
flextable(watergap_stats)

remove(watergap_stats)
gc()

Environmental distribution of reference streamgauging stations

loadd(envhist)
grid.newpage() 
grid.draw(envhist)


pdf(paste0(file.path(resdir, 'figures//envhist_'), 
           format(Sys.Date(), '%Y%m%d'), '.pdf'), 
    width=10, height=9)
grid.draw(envhist)
dev.off()

remove(envhist)
gc()

Methods - Table 2. Specification and benchmark comparison of models

loadd(tablebm_classif1_u10)
loadd(tablebm_regr1_u10)
loadd(tablebm_classif2_u10)
print('Setup table u10')
setup_table_u10 <- rbindlist(list(tablebm_classif1_u10$setup,
                              tablebm_regr1_u10$setup,
                              tablebm_classif2_u10$setup),
                         fill=T, use.names=T)

flextable(setup_table_u10)

print('Results table u10')
results_table_u10 <- rbindlist(list(tablebm_classif1_u10$results,
                                tablebm_regr1_u10$results,
                                tablebm_classif2_u10$results),
                           fill=T, use.names=T)
flextable(results_table_u10)

print('Setup table o1')
loadd(tablebm_classif1_o1)
loadd(tablebm_regr1_o1)
loadd(tablebm_classif2_o1)
setup_table_o1 <- rbindlist(list(tablebm_classif1_o1$setup,
                              tablebm_regr1_o1$setup,
                              tablebm_classif2_o1$setup),
                         fill=T, use.names=T)

flextable(setup_table_o1)

print('Results table o1')
results_table_o1 <- rbindlist(list(tablebm_classif1_o1$results,
                              tablebm_regr1_o1$results,
                              tablebm_classif2_o1$results),
                         fill=T, use.names=T)
flextable(results_table_o1)

remove(tablebm_classif1_u10)
remove(tablebm_regr1_u10)
remove(tablebm_classif2_u10)
gc()

Methods - Figure 2. Benchmark comparison of models through curves

loadd(misclass_plot_u10)
print(misclass_plot_u10)

loadd(misclass_plot_o1)
print(misclass_plot_o1)


gpredsdt[, mlr3measures::bacc(factor(as.character(intermittent_o1800),
                                      levels=c('0', '1')),
                               factor(IRpredcat_CVsp, levels=c('0','1')))]
gpredsdt[, mlr3measures::bacc(factor(as.character(intermittent_o1800),
                                      levels=c('0', '1')),
                               factor(IRpredcat_CVnosp, levels=c('0','1')))]

Main text - Figure 2. Variable importance for top 20 variables

loadd(vimp_plot_u10)
loadd(vimp_plot_o1)

fwrite(vimp_plot_u10$data[, .(varnames, imp_wmean, imp_wsd, Keyscale, Keystat, 
                              Category, Attribute, Source, Citation, varname)],
       paste0(file.path(resdir, 'figures//vimpdata_u10_'), 
           format(Sys.Date(), '%Y%m%d'), '.csv')
)

fwrite(vimp_plot_o1$data[, .(varnames, imp_wmean, imp_wsd, Keyscale, Keystat, 
                              Category, Attribute, Source, Citation, varname)],
       paste0(file.path(resdir, 'figures//vimpdata_o1_'), 
           format(Sys.Date(), '%Y%m%d'), '.csv')
)

p_u10 <- vimp_plot_u10 +
  theme(plot.margin = unit(c(10,260,0,0), "pt")) +
  labs(x=bquote('A. Gauges with mean annual discharge < 10'~m^3~s^-1)) + 
  theme(plot.margin = unit(c(0,0,0,0), 'cm'))

p_o1format <- vimp_plot_o1 +
  coord_flip(ylim=c(min(vimp_plot_o1$data[, max(imp_wmean+imp_wsd)+1], 100), 0),
             clip='off') +
  scale_x_discrete(name=expression(
    B.~Gauges~with~mean~annual~discharge >= 1~m^3~s^-1),
    labels = function(x) {
      stringr::str_wrap(tolower(x), width = 25)
    },
    limits = rev,
    position='top') +
  scale_y_reverse(position='right', expand=c(0,0)) +
  theme(legend.position = 'none',
        plot.background = element_blank(),
        panel.background = element_blank(),
        plot.margin = unit(c(0,0,0,0.1), 'cm'))

pdf(paste0(file.path(resdir, 'figures//vimp_'), 
           format(Sys.Date(), '%Y%m%d'), '.pdf'), 
    width=7.5, height=6)
p_u10 + p_o1format
dev.off()

remove(vimp_plot_u10)
remove(vimp_plot_o1)
gc()

Main text - Figure 3. Partial dependence plots

loadd(pd_plot_u10)
lapply(pd_plot_u10, function(p) plot(p))

loadd(pd_plot_o1)
lapply(pd_plot_o1, function(p) plot(p))

remove(pd_plot_u10)
remove(pd_plot_o1)
gc()

Methods - Figure 3 A. Predictions uncertainty by metavariable and environment

loadd(gaugeIPR_plot)
grid.newpage()
grid.draw(gaugeIPR_plot)

remove(gaugeIPR_plot)
gc()

Final binned summary statistics for split model approach(< 10 and >= 1)

loadd(bin_finalmisclass_IRpredcat_CVnosp)
loadd(bin_finalmisclass_IRpredcat_CVsp)
loadd(gpredsdt)

print('Split model approach: 3-fold non-spatial CV')
print(paste0(
  'Overal bacc:', 
  gpredsdt[, round(mlr3measures::bacc(intermittent_o1800, 
                                      as.factor(IRpredcat_CVnosp)),
                   2)]
)
)
flextable(bin_finalmisclass_IRpredcat_CVnosp)

print('Slit model approach: 40-fold spatial CV')
print(paste0(
  'Overal bacc:', 
  gpredsdt[, round(mlr3measures::bacc(intermittent_o1800, 
                                      as.factor(IRpredcat_CVsp)),
                   2)]
)
)
flextable(bin_finalmisclass_IRpredcat_CVsp)

remove(bin_finalmisclass_IRpredcat_CVnosp)
remove(bin_finalmisclass_IRpredcat_CVsp)

Binned summary statistics for single model approach

loadd(res_featsel_cv_all)
print('Single model approach: 3-fold non-♣spatial CV')
bin_misclass_all <- bin_misclass(in_resampleresult = res_featsel_cv_all,
                                 binvar = 'dis_m3_pyr',
                                 binfunc = 'manual',
                                 binarg = c(0.1, 1, 10, 100, 1000, 10000, 1000000),
                                 interthresh=0.5,
                                 spatial_rsp=FALSE
)
flextable(bin_misclass_all)

remove(res_featsel_cv_all)
gc()

Basin-level accuracy and bias plots

loadd(basinBACC)

png(paste0(file.path(resdir, 'figures//basaccbias_'), 
           format(Sys.Date(), '%Y%m%d'), '.png'), 
    width=6, height=3.5, units='in', res=1000)
basinBACC$plot_acc + 
  basinBACC$plot_bias + theme(legend.position = 'none') +
  plot_annotation(tag_levels = 'a') & 
  theme(plot.tag = element_text(size = 12))
dev.off()

png(paste0(file.path(resdir, 'figures//jcdeviate_'), 
           format(Sys.Date(), '%Y%m%d'), '.png'), 
    width=5, height=5, units='in', res=1000)
basinBACC$plot_jcdeviate
dev.off()

gc()

Comparison of results with regional-national estimates

getdisplay_totalest <- function(target, subtarget='data', minbin=2) {
  flextable(target[[subtarget]] %>%
              rbind(.[bin>=minbin, list(bin='world',
                                  perc = weighted.mean(perc, binsumlength)
              ), by=dat],fill=T)
  )
}

loadd(fr_plot)
loadd(us_plot)
loadd(au_plot)
au_plot <- au_plot$plot + theme(axis.title.y = element_blank(), 
                                legend.title=element_blank()) + 
  labs(x=bquote('Surface area'~km^2))

pdf(paste0(file.path(resdir, 'figures//comparisonhist_'), 
           format(Sys.Date(), '%Y%m%d'), '.pdf'), 
    width=5, height=11)
grid.arrange(fr_plot$plot, us_plot$plot, au_plot, left = 'Prevalence of intermittency (% river length)')
dev.off()

print("Data for comparison: US - all")
getdisplay_totalest(us_plot, subtarget='data_all')

print("Data for comparison: US - no artificial")
getdisplay_totalest(us_plot, subtarget='data_noartificial')

print("Data for comparison: France")
getdisplay_totalest(fr_plot)

print("Data for comparison: Australia")
getdisplay_totalest(au_plot, minbin=1)

remove(ft_plot)
remove(us_plot)
remove(au_plot)
gc()

loadd(fr_plot_mdur30)
loadd(us_plot_mdur30)
loadd(au_plot_mdur30) 
au_plot_mdur30 <- au_plot_mdur30$plot + theme(axis.title.y = element_blank(), 
                                              legend.title=element_blank()) + 
    labs(x=bquote('Surface area'~km^2))
pdf(paste0(file.path(resdir, 'figures//comparisonhist_mdur30_'), 
           format(Sys.Date(), '%Y%m%d'), '.pdf'), 
    width=5, height=11)
grid.arrange(fr_plot_mdur30$plot, us_plot_mdur30$plot, au_plot_mdur30, 
             left = 'Prevalence of intermittency (% river length)') 
dev.off()

print("Data for comparison, mdur >= 30: US - all")
getdisplay_totalest(us_plot_mdur30, subtarget='data_all')

print("Data for comparison, mdur >= 30: US - no artificial")
getdisplay_totalest(us_plot_mdur30, subtarget='data_noartificial')

print("Data for comparison, mdur >= 30: France")
getdisplay_totalest(fr_plot_mdur30)

print("Data for comparison, mdur >= 30: Australia")
getdisplay_totalest(au_plot_mdur30, minbin=1)

remove(ft_plot_mdur30)
remove(us_plot_mdur30)
remove(au_plot_mdur30)
gc()

Comparison of results with on-the-ground observations for PROSPER

loadd(pnw_plot)
print(pnw_plot$plot)

pnw_measures <- pnw_plot$stats
print(paste0('Balanced accuracy of predictions based on PROSPER: ',
             round(pnw_measures$bacc, 3)))
print(paste0('AUC based on PROSPER: ',
             round(pnw_measures$auc, 3)))
print(paste0('Misclassification rate based on PROSPER: ',
             round(pnw_measures$ce, 3)))
print(paste0('Sensitivity based on PROSPER: ',
             round(pnw_measures$sen, 3)))
print(paste0('Specificity rate based on PROSPER: ',
             round(pnw_measures$spe, 3)))


print(paste0('Total number of obs: ',
             pnw_measures$nobs_total
))
print(paste0('Total number of reaches: ',
             pnw_measures$nreaches
))
print(paste0('Total number of perennial reaches: ',
              pnw_measures$nreaches_perennial
))
print(paste0('Total number of non-perennial reaches: ',
              pnw_measures$nreaches_nonperennial
))

remove(pnw_plot)
gc()

Comparison of results with on-the-ground observations for ONDE

loadd(onde_plot)
print(onde_plot$plot)

onde_measures <- onde_plot$stats
print(paste0('Balanced accuracy of predictions based on ONDE: ',
             round(onde_measures$bacc, 3)))
print(paste0('AUC based on ONDE: ',
             round(onde_measures$auc, 3)))
print(paste0('Misclassification rate based on ONDE: ',
             round(onde_measures$ce, 3)))
print(paste0('Sensitivity based on ONDE: ',
             round(onde_measures$sen, 3)))
print(paste0('Specificity rate based on ONDE: ',
             round(onde_measures$spe, 3)))

print(paste0('Total number of obs: ',
             onde_measures$nobs_total
))
print(paste0('Total number of reaches: ',
             onde_measures$nreaches
))
print(paste0('Total number of perennial reaches: ',
              onde_measures$nreaches_perennial
))
print(paste0('Total number of non-perennial reaches: ',
              onde_measures$nreaches_nonperennial
))

remove(onde_plot)
gc()

Environmental variables used in model training + variable selection results

loadd(predvars)

#Write predvars to csv
fwrite(predvars, file.path(resdir, paste0('predictor_variables.csv')))

#Make table
flextable(predvars[,c('Category', 'Attribute', 'Spatial representation',
                      'Temporal/Statistical aggreg.', 'Source', 'Citation'),
                   with=F])

remove(predvars)
gc()

Tables of intermittence by categories for base model

loadd(gaugestats_format)
get_tabletarget <- function(in_gaugeformat) {
  lapply(names(in_gaugeformat), function(var) {
    targetname <- paste0('globaltables_', var)
    #print(targetname)
    tryCatch(return(readd(eval(targetname), character_only = T)$`0.5` %>%
                      setDT(tab) %>%
                      setorder(-`Total stream length (10^3 km)`)),
             error=function(e) NULL)
  })
}

tables <- get_tabletarget(gaugestats_format) %>%
  compact

ft <- list()
for (i in 1:length(tables)) {
  numcols <- names(tables[[i]])[tables[[i]][,sapply(.SD, class)=='numeric']]
  ft[[i]] <- flextable(tables[[i]][, (numcols) := lapply(.SD, function(x) as.integer(round(x))),
                              .SDcols = numcols]) %>%
    autofit
}
ft[[1]]
ft[[2]]
ft[[3]]
ft[[4]]

print('Extrapolation')
loadd(globaltable_clzextend)
flextable(globaltable_clzextend)

remove(gaugestats_format)
gc()

Tables of intermittence by categories for mdur 30 model

loadd(gaugestats_format)
get_tabletarget_mdur30 <- function(in_gaugeformat) {
  lapply(names(in_gaugeformat), function(var) {
    targetname <- paste0('globaltables_', var, '_mdur30')
    #print(targetname)
    tryCatch(return(readd(eval(targetname), character_only = T)$`0.5` %>%
                      setDT(tab) %>%
                      setorder(-`Total stream length (10^3 km)`)),
             error=function(e) NULL)
  })
}

tables <- get_tabletarget_mdur30(gaugestats_format) %>%
  compact

ft_mdur30 <- list()
for (i in 1:length(tables)) {
  numcols <- names(tables[[i]])[tables[[i]][,sapply(.SD, class)=='numeric']]
  ft_mdur30[[i]] <- flextable(tables[[i]][, (numcols) := lapply(.SD, function(x) as.integer(round(x))),
                              .SDcols = numcols]) %>%
    autofit
}

print(paste0('Table organized by: ', ft[[1]]$header$col_keys[1]))
ft_mdur30[[1]]
print(paste0('Table organized by: ', ft[[2]]$header$col_keys[1]))
ft_mdur30[[2]]
print(paste0('Table organized by: ', ft[[3]]$header$col_keys[1]))
ft_mdur30[[3]]
print(paste0('Table organized by: ', ft[[4]]$header$col_keys[1]))
ft_mdur30[[4]]

print('Extrapolation')
loadd(globaltable_clzextend_mdur30)
flextable(globaltable_clzextend_mdur30)

remove(gaugestats_format)
gc()

Statistics for press

loadd(IRESextra)

#% IRES for MAF >= 0.1 m3/s in NZ
IRESextra$preds[substr(PFAF_IDclz, 1, 2) == '57', list(cumLpred = sum(cumL_pred),
                                                       cumLref = sum(cumL_cutoffref),
                                                       cumIRES_conservative = weighted.mean(percinter_all_conservative, cumL_pred),
                                                       cumIRES_GAM = weighted.mean(percinter_all_GAM, cumL_pred, na.rm=T)
)]

#% IRES for MAF >= 0.1 m3/s in Amazon basin
IRESextra$preds[substr(PFAF_IDclz, 1, 2) == '62', list(cumLpred = sum(cumL_pred),
                                                       cumLref = sum(cumL_cutoffref),
                                                       cumIRES_conservative = weighted.mean(percinter_all_conservative, cumL_pred),
                                                       cumIRES_GAM = weighted.mean(percinter_all_GAM, cumL_pred, na.rm=T)
)]

#% IRES for MAF >= 0.1 m3/s in Europe
IRESextra$preds[(substr(PFAF_IDclz, 1, 2) %in% as.character(seq(21,25)))
                & !(substr(PFAF_IDclz, 4, 5) %in% c('1','2','14')), 
                list(cumLpred = sum(cumL_pred),
                     cumLref = sum(cumL_cutoffref),
                     cumIRES_conservative = weighted.mean(percinter_all_conservative, cumL_pred),
                     cumIRES_GAM = weighted.mean(percinter_all_GAM, cumL_pred, na.rm=T)
                )]

NaiaraLopezRojo/globalIRmap documentation built on Dec. 17, 2021, 5:19 a.m.

rdrr.io home R language documentation Run R code online

CRAN packages Bioconductor packages R-Forge packages GitHub packages

Note that we can't provide technical support on individual packages. You should contact the package authors for that.

NaiaraLopezRojo/globalIRmap
What the Package Does (One Line, Title Case)

In NaiaraLopezRojo/globalIRmap: What the Package Does (One Line, Title Case)

Geographical distribution of reference streamgauging stations

Statistics for paper

Threshold sensitivity

IRES extrapolation

Gauges map

WaterGAP stats

Environmental distribution of reference streamgauging stations

Methods - Table 2. Specification and benchmark comparison of models

Methods - Figure 2. Benchmark comparison of models through curves

Main text - Figure 2. Variable importance for top 20 variables

Main text - Figure 3. Partial dependence plots

Methods - Figure 3 A. Predictions uncertainty by metavariable and environment

Final binned summary statistics for split model approach(< 10 and >= 1)

Binned summary statistics for single model approach

Basin-level accuracy and bias plots

Comparison of results with regional-national estimates

Comparison of results with on-the-ground observations for PROSPER

Comparison of results with on-the-ground observations for ONDE

Environmental variables used in model training + variable selection results

Tables of intermittence by categories for base model

Tables of intermittence by categories for mdur 30 model

Statistics for press

R Package Documentation

Browse R Packages

We want your feedback!

NaiaraLopezRojo/globalIRmap What the Package Does (One Line, Title Case)

In NaiaraLopezRojo/globalIRmap: What the Package Does (One Line, Title Case)

Geographical distribution of reference streamgauging stations

Statistics for paper

Threshold sensitivity

IRES extrapolation

Gauges map

WaterGAP stats

Environmental distribution of reference streamgauging stations

Methods - Table 2. Specification and benchmark comparison of models

Methods - Figure 2. Benchmark comparison of models through curves

Main text - Figure 2. Variable importance for top 20 variables

Main text - Figure 3. Partial dependence plots

Methods - Figure 3 A. Predictions uncertainty by metavariable and environment

Final binned summary statistics for split model approach(< 10 and >= 1)

Binned summary statistics for single model approach

Basin-level accuracy and bias plots

Comparison of results with regional-national estimates

Comparison of results with on-the-ground observations for PROSPER

Comparison of results with on-the-ground observations for ONDE

Environmental variables used in model training + variable selection results

Tables of intermittence by categories for base model

Tables of intermittence by categories for mdur 30 model

Statistics for press

R Package Documentation

Browse R Packages

We want your feedback!

NaiaraLopezRojo/globalIRmap
What the Package Does (One Line, Title Case)