extrapolate_networklength: Extrapolate river network length

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

Description

Train models and use them to extrapolate the global length of rivers.

Usage

extrapolate_networklength(
  inp_riveratlas,
  min_cutoff = 0.1,
  dispred = seq(0.01, 0.09, 0.01),
  interactive = T,
  grouping_var = "PFAF_IDclz"
)

Arguments

inp_riveratlas	full path to RiverATLAS (v1.0) attribute table.
min_cutoff	(numeric) minimum discharge to include in model training.
dispred	(numeric vector) discharge values for which to produce model predictions.
interactive	(logical) whether to print results and make plots for user to interactively evaluate results and troubleshoot.
grouping_var	variable with which to subset the dataset into groups. A separate model is trained on each group.

Details

The prevalence of IRES was independently extrapolated for a total of 465 spatial sub-units representing all occurring intersections of 62 river basin regions (BasinATLAS level 2 subdivisions) and 18 climate zones (Global Environmental Stratification). For each basin–climate sub-unit, we first extrapolated the empirical cumulative distribution of total stream length (of all reaches with MAF ≥ 0.1 m3 s−1) down to 0.01 m3 s−1 MAF using a generalized additive model (GAM). We excluded reaches larger than the 95th percentile of MAF (that is, the largest rivers) within the sub-unit from model fitting to avoid common discontinuities at the high end of the empirical distribution that can affect the low end of the power-law-like trendline.

Value

list containing a data.table and two plots. The data.table contains an estimate stream length for each discharge size class and climate-basin sub-unit. cumL_pred is the cumulative length of all rivers and streams with MAF > dispred. cumL_cutoffref is the cumulative river length at MAF == min_cutoff. cumL_predextra = cumL_pred - cumL_cutoffref.

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