tidysoilwater

Tidy tools for soil water retention and unsaturated hydraulic conductivity modelling. Pipe-compatible functions that accept and return tibbles, with tidy evaluation support for column arguments. Implements the Van Genuchten (1980) / Mualem (1976) model family with parameter fitting from measured (h, θ) data.

Installation

# Install from GitHub:
pak::pak("Taakefyrsten/tidysoilwater")

Quick start

library(tidysoilwater)
library(tibble)

# Soil water retention curve for a loam soil
soil <- tibble(h = c(0, 10, 33, 100, 330, 1000, 5000, 15000))

swrc_van_genuchten(soil,
  theta_r = 0.065, theta_s = 0.41,
  alpha = 0.075, n = 1.89, h = h)

# Fit parameters from observed data
observed <- tibble(
  h     = c(0, 10, 100, 1000, 5000, 15000),
  theta = c(0.44, 0.40, 0.32, 0.20, 0.12, 0.08)
)

fit_swrc(observed, theta_col = theta, h_col = h)

Multi-pedon workflows

Group your data with dplyr::group_by() and fit_swrc() fits each group independently. Use workers for parallel fitting on Unix:

library(dplyr)

pedons |>
  group_by(pedon_id) |>
  fit_swrc(theta_col = theta, h_col = h,
           workers = parallel::detectCores(logical = FALSE))

Performance

Task	Time
SWRC for 1 000 000 rows	< 20 ms
K(h) for 1 000 000 rows	< 40 ms
Fit 500 pedons (4 cores)	< 2 s

Learn more

• vignette("soil-water-retention") — Van Genuchten model parameters
• vignette("hydraulic-conductivity") — Mualem-Van Genuchten K(h)
• vignette("fitting-diagnostics") — convergence, residuals, uncertainty
• vignette("pedotransfer") — PTF-based parameter estimation from texture
• vignette("multi-pedon-workflow") — large-scale parallel fitting

Full documentation: https://taakefyrsten.github.io/tidysoilwater