{infiltrodiscR}

Soil Biophysics Lab

Sara Acevedo / Carolina Giraldo

Hello!

Sara Acevedo 👩‍💻 Ph.D. Engineering Sciences
Carolina Giraldo 👩‍💻 Ph.D. (c) Engineering Sciences
Soil Biophysics Lab
Pontificia Universidad Católica de Chile

InfiltrodiscR

In this session:

Data wrangling
Functions
Calculating K
Plotting

R code and materials

The presentations contain actual R code (can be copied and pasted directly). This repository will contain the material and content for this session.

1 + 1 # copy and paste!

[1] 2

The repository is here:

Github repo

Requirements

`infiltrodiscR::` needs:

Tidy oriented package (you need to install {tidyverse})
Your data must be in two numeric columns: time and infiltration
You need data on soil texture (USDA) and suction (like “xcm”) for each measurement

How to install it

install.packages("devtools")
install.packages("tidyverse")
install.packages("broom")
install.packages("glue")

devtools::install_github("biofisicasuelos/infiltrodiscR")

`infiltrodiscR::` functions

infiltration() This function calculates cumulative infiltration and the square root of time, using time and volume recorded.

vg_par This function returns the parameter A, no_h and alpha related to the van Genuchten parameters (van Genuchten, 1980), from tabulated data calculated for a radius of 2.25 cm, including 12 soil texture classes and suctions from -0.5 cm to -7 cm. Data from (Decagon Devices, Inc., 2005)

`infiltrodiscR::` functions

parameter_A This function returns the parameter A calculated from the equation proposed by Zhang (1997)

Tutorial part 1: data wrangling

Let’s load some data

library(tidyverse)
library(fs)
library(kableExtra)

info_soils <- read_csv("data_suelos/info_suelos.csv")

csv_data <- dir_ls("data", regexp = "\\.csv$")

data_infiltration <- 
csv_data %>% 
  map_dfr(read_csv, .id = "source") %>% 
  mutate(source = str_sub(source, start = 12, end = -5)) %>% 
  rename(soil = source)

Tutorial part 1: data wrangling

Three columns: soil ID, time and volume

data_infiltration

# A tibble: 23 × 3
   soil   time volume
   <chr> <dbl>  <dbl>
 1 a         0     93
 2 a         5     89
 3 a        10     86
 4 a        15     83
 5 a        20     80
 6 a        25     77
 7 a        30     75
 8 a        35     73
 9 b         0     95
10 b        10     91
# ℹ 13 more rows

Tutorial part 1: data wrangling

Now the infiltration data will be nested:

data_infiltration_nested <- 
data_infiltration %>% 
  group_by(soil) %>% 
  nest()

Each soil (or measurement or location) has its infiltration data nested

Tutorial part 1: data wrangling

Now we need information about soil texture (USDA) and tension

data_infiltration_all <- 
data_infiltration_nested %>% 
  left_join(info_soils, by = join_by(soil == suelo))

data_infiltration_all

# A tibble: 3 × 7
# Groups:   soil [3]
  soil  data              sand  silt  clay suction texture   
  <chr> <list>           <dbl> <dbl> <dbl> <chr>   <chr>     
1 a     <tibble [8 × 2]>    40    40    20 2cm     silt loam 
2 b     <tibble [8 × 2]>    20    30    50 2cm     clay      
3 c     <tibble [7 × 2]>    70    20    10 3cm     sandy loam

Tutorial part 2: functions

First, we use the infiltration() function

library(infiltrodiscR)

data_infiltration_all %>% 
  mutate(data = map(data, ~ infiltration(.), data = .x)) %>% 
  unnest(data) # let's check the data nested created

# A tibble: 23 × 11
# Groups:   soil [3]
   soil   time volume sqrt_time volume_infiltrated infiltration  sand  silt
   <chr> <dbl>  <dbl>     <dbl>              <dbl>        <dbl> <dbl> <dbl>
 1 a         0     93      0                     0         0       40    40
 2 a         5     89      2.24                  4         0.25    40    40
 3 a        10     86      3.16                  7         0.44    40    40
 4 a        15     83      3.87                 10         0.63    40    40
 5 a        20     80      4.47                 13         0.82    40    40
 6 a        25     77      5                    16         1.01    40    40
 7 a        30     75      5.48                 18         1.13    40    40
 8 a        35     73      5.92                 20         1.26    40    40
 9 b         0     95      0                     0         0       20    30
10 b        10     91      3.16                  4         0.25    20    30
# ℹ 13 more rows
# ℹ 3 more variables: clay <dbl>, suction <chr>, texture <chr>

Tutorial part 2: functions

Let’s check the var_par() function

data_infiltration_all %>% 
  select(-sand,-silt,-clay) %>% 
  vg_par()

# A tibble: 3 × 7
# Groups:   soil [3]
  soil  data             suction texture    alpha  n_ho value_A
  <chr> <list>           <chr>   <chr>      <dbl> <dbl>   <dbl>
1 a     <tibble [8 × 2]> 2cm     silt loam  0.02   1.41    7.93
2 b     <tibble [8 × 2]> 2cm     clay       0.008  1.09    4.30
3 c     <tibble [7 × 2]> 3cm     sandy loam 0.075  1.89    3.93

Tutorial part 3: calculating K

The hydraulic conductivity of the soil K at a specific suctions is calculated as: K(h) = C1 / A. Parameter C1 is calculated fitting a polynomial function of the second degree (y = ax2+b), where a is parameter C1, x is the square root of time and y is the cumulative infiltration. For this step, we use the package broom and base R.

Tutorial part 3: calculating K

Here we use lm(poly) to get a polynomial regression

library(broom)

data_infiltration_all %>% 
  mutate(data = map(data, ~ infiltration(.), data = .x)) %>% 
  vg_par() %>% 
    mutate(
    fit = map(data, ~ lm(infiltration ~ poly(sqrt_time, 2, raw = TRUE), data = .x)),
    tidied = map(fit, tidy)
  )

# A tibble: 3 × 12
# Groups:   soil [3]
  soil  data      sand  silt  clay suction texture    alpha  n_ho value_A fit   
  <chr> <list>   <dbl> <dbl> <dbl> <chr>   <chr>      <dbl> <dbl>   <dbl> <list>
1 a     <tibble>    40    40    20 2cm     silt loam  0.02   1.41    7.93 <lm>  
2 b     <tibble>    20    30    50 2cm     clay       0.008  1.09    4.30 <lm>  
3 c     <tibble>    70    20    10 3cm     sandy loam 0.075  1.89    3.93 <lm>  
# ℹ 1 more variable: tidied <list>

Tutorial part 3: calculating K

The coefficients are unnested and only “a” (or C1) are filtered

data_infiltration_all %>% 
  mutate(data = map(data, ~ infiltration(.), data = .x)) %>% 
  vg_par() %>% 
    mutate(
    fit = map(data, ~ lm(infiltration ~ poly(sqrt_time, 2, raw = TRUE), data = .x)),
    tidied = map(fit, tidy)
  ) %>% 
  unnest(tidied) %>% 
  filter(term == "poly(sqrt_time, 2, raw = TRUE)2")  # to filter "a" which = C1

# A tibble: 3 × 16
# Groups:   soil [3]
  soil  data      sand  silt  clay suction texture    alpha  n_ho value_A fit   
  <chr> <list>   <dbl> <dbl> <dbl> <chr>   <chr>      <dbl> <dbl>   <dbl> <list>
1 a     <tibble>    40    40    20 2cm     silt loam  0.02   1.41    7.93 <lm>  
2 b     <tibble>    20    30    50 2cm     clay       0.008  1.09    4.30 <lm>  
3 c     <tibble>    70    20    10 3cm     sandy loam 0.075  1.89    3.93 <lm>  
# ℹ 5 more variables: term <chr>, estimate <dbl>, std.error <dbl>,
#   statistic <dbl>, p.value <dbl>

Tutorial part 3: calculating K

final_calculations <-
data_infiltration_all %>% 
  mutate(data = map(data, ~ infiltration(.), data = .x)) %>% 
  vg_par() %>% 
    mutate(
    fit = map(data, ~ lm(infiltration ~ poly(sqrt_time, 2, raw = TRUE), data = .x)),
    tidied = map(fit, tidy)
  ) %>% 
  unnest(tidied) %>% 
  filter(term == "poly(sqrt_time, 2, raw = TRUE)2") %>%   # to filter "a" which = C1
  rename(C1 = estimate) %>%
  parameter_A() %>% # function to calculate A (not from the table)
  mutate(K_h = C1 / parameter_A) # finally the value of K!

Tutorial part 3: calculating K

final_calculations %>% 
  select(soil, suction, texture, parameter_A, K_h)

# A tibble: 3 × 5
# Groups:   soil [3]
  soil  suction texture    parameter_A       K_h
  <chr> <chr>   <chr>            <dbl>     <dbl>
1 a     2cm     silt loam         7.93  0.00331 
2 b     2cm     clay              4.30  0.00305 
3 c     3cm     sandy loam        3.93 -0.000784

Tutorial part 4: plotting

Because data is nested, it is possible also iterate plots

library(glue)

plots <- 
final_calculations %>%
  mutate(plot = map2(
    data,
    soil,
    ~ ggplot(
      data = .x,
      aes(
        x = sqrt_time,
        y = infiltration
      )
    ) +
      geom_point() +
      stat_smooth(method = 'lm', formula = y ~ poly(x, 2)) +
      ggtitle(glue(
        "Soil : {soil}"
      ))
  ))

Tutorial part 4: plotting

Because data is nested, it is possible also iterate plots

patchwork::wrap_plots(plots$plot, ncol = 3)

A lot of code for today! 🧠 🔥

{infiltrodiscR}

Hello!

InfiltrodiscR

R code and materials

Requirements

infiltrodiscR:: needs:

How to install it

infiltrodiscR:: functions

infiltrodiscR:: functions

Tutorial part 1: data wrangling

Tutorial part 1: data wrangling

Tutorial part 1: data wrangling

Tutorial part 1: data wrangling

Tutorial part 2: functions

Tutorial part 2: functions

Tutorial part 3: calculating K

Tutorial part 3: calculating K

Tutorial part 3: calculating K

Tutorial part 3: calculating K

Tutorial part 3: calculating K

Tutorial part 4: plotting

Tutorial part 4: plotting

A lot of code for today! 🧠 🔥

`infiltrodiscR::` needs:

`infiltrodiscR::` functions

`infiltrodiscR::` functions