Leveraging in situ sensors for soil water release curves

This repository hosts time-series of soil data (VWC and water potential) provided by Joel Biederman and Fangyue Zhang to address how to derive soil moisture release curves from paired in situ sensors of volumetric water content (VWC) and soil water potential (SWP). The incubator proposal describes the initial scope of work, which expanded to include the following:

(1) explore whether daily or half-hourly data was more suitable for moisture release curves

(2) interactively visualize the data in a Shiny app

(3) model the VWC-SWP relationship and derive Van Genuchten parameters

(4) create a function with a look up table to convert in VWC into SWP for RainMan soils

Data description

There are about 50 VWC sensors and 6 SWP sensors installed. The 6 SWP sensors are matched with 6 SWC sensors in 2 plots and at 3 depths in profile, for the S1 and S4 treatments. Soil temperature is also provided for screening, correcting, or interpreting diurnal or temperature induced oscillations, which mostly occur in the SWP data at dry values.

Key dates include:

• April/May 2019, plot construction and installation of most VWC sensors
• August 2019, SWP sensors installed
• 10/1/2019, start of data provided
• 10/31/2019, large irrigation applied to prep for transplant
• 11/1/2019, 30 Digitaria californica (Arizona cottontop) transplanted to plot. Plots then receive natural rainfall + irrigation every 1-2 weeks
• May 2020, irrigation ceased and roofs installed
• 7/13/2020, irrigation treatments (temporal repackaging of summer precipitation, S1-S4) applied. Each house contains 3 replicates of each of 4 summer treatments (12 plots per house, 5 houses)
• 10/1/2020, separate winter rainfall treatments ended and all plots have received equal winter treatments, applied every 2 weeks

Repository description

• data/ contains csv files provided by Fangyue

  • Daily_Rain_Gage.csv and Daily_Raingage_02262022.csv contain daily total precip from an on-site gauge
  • Halfhourly_TS_011321.csv and Halfhourly_TS_112721.csv contain the soil temperature starting from fall 2019
  • Halfhourly_VWC_SWP_Weather_112721.csv and Halfhourly_VWC_Weather_030621.csv contain the soil moisture, soil water potential, and weather variables from fall 2019
  • irrigation.csv contains the amount and date of irrigation provided to each of the 4 treatments: S1, S2, S3, S4
  • treatments.csv contains the summer and/or winter treatment of each house and plot

• models/ are the non-linear Bayesian models to fit data to Van Genuchten function

  • v1/ contains the non-hierarhical model, scripts, and report
    • out.Rdata daily data created from scripts/05_select_data.R
    • 01_run_model.R runs model
    • 02_plot_output.R plots output
    • VG.jags JAGS model, separate parameters for S1 and S4
    • model_report.Rmd reports model description and output
  • v2/ contains the hierarchical model, scripts, and report
    • out.Rdata daily data created from scripts/05_select_data.R
    • 01_run_model.R runs model
    • 02_plot_output.R plots output
    • 03_create_table.R creates lookup table with posterior parameters
    • VG_hierarchy.jags JAGS model, estimate treatment- and site-level parameters
    • model_report.Rmd describes model, output, equation, and next steps

• plots/ containts static figures

  • 3_4_byseason_daily.jpg, 3_7_byseason_daily.jpg plot time series and bivariate plots of daily VWC and SWP by depth and season for the S1 and S4 treatments, respectively, generated by scripts/01_exploration_daily.Rmd
  • VG_curves/S1_shallow_byseason.png plots the SWP-VWC relationship in the shallow S1 treatment for different periods of time, generated by scripts/04_explore_pulses.R
  • weekly/ plots time series and bivariate plots of halfhourly VWC and SWP by depth and treatment for each week, generated by scripts/02_exploration_halfhourly.Rmd

• scripts/

  • 01_exploration_daily.Rmd reads in, organizes, and plots daily data
  • 02_exploration_halfhourly.Rmd reads in, organizes, and plots half-hourly data
  • 03_organize_daily.R prepares daily data as .Rdata files saved to soildata_app/
  • 04_explores_pulses.R explores criteria for selecting VWC and SWP data for the Van Genuchten curves
  • 05_select_data.R selects data and saves to models/v1/ and models/v2/

• soildata_app/ contains the code for the Shiny app, documented here

  • app2.R that creates an interactive visualization with 3 tabs
  • The following .Rdata files were created in scripts/03_organize.daily.Rmd
    • soilTempDaily.Rdata daily soil temperature by plot and depth
    • soilWaterContentDaily.Rdata daily VWC by plot and depth
    • airTempDaily.Rdata daily air temperature, RH, and VPD both inside and outside the shelters
    • comboDaily.Rdata appreciation and/or irrigation (mm) in each treatment
    • season.Rdata start and end dates of each water year season
    • soilWPWCDaily.Rdata matched VWC and SWP data for the S1 and S4 plot in House 3

• source/ contains custom R functions

  • lookup.Rdata contains the output of models/v2/03_create_table.R as a lookup table for converting VWC to SWP
  • swc2swp.R uses lookup.Rdata to convert VWC to SWP based on either site, S1, or S4 parameters and returns either median, 0.25 percentile, or 0.75 percentile SWP