Soil Data Aggregation using R

This document is based on:

• aqp version 1.8-3
• soilDB version 1.5-3
• sharpshootR version 0.7-2

• an analysis platform: calculator, statistics program, GIS, etc…
• R syntax: vocabulary to explore, summarize, and model data

• an integrator: analysis + GIS + database connectivity
• ODBC and GDAL link R to nearly all possible formats/interfaces

• Base R (functionality is extended through packages)

  • basic summaries of quantitative or qualitative data
  • data exploration via graphics
  • GIS data processing and analysis

• Integrated Workflow (self-documenting, repeatable)

  • report writing: code + summaries in a single document
  • digital soil mapping: modeling properties and classes
  • evaluation of OSD, KSSL, NASIS, … for SDJR or related

• Soil Science Specific Packages

  • visualization, aggregation, classification → aqp package
  • access to commonly used soil databases → soilDB package
  • misc. soil survey specific functions → sharpshootR package

• Be sure to follow embedded links to full documentation
• Check out the AQP project website

• Browse some of the various documentation on R

• The first time you open RStudio, install required packages:

install.packages('aqp', dep=TRUE) 
install.packages("RODBC", dep=TRUE)
install.packages('soilDB', dep=TRUE)

• Every subsequent time, load required packages:

library(aqp)
library(soilDB)
library(latticeExtra)
# ... other libraries you might be using

• connection allows R to pull data from a NASIS selected set via the soilDB package
• connection sets up a small file that stores the NASIS login and password

• this is a one-time operation → ODBC setup instructions

• getting data from NASIS is as simple as:

# get pedons from the selected set
pedons <- fetchNASIS()

# plot the first 15
plot(pedons[1:15, ])

# get component data from the selected set
components <- get_component_data_from_NASIS_db()

• be sure to read the documentation → fetchNASIS()

1. Select a set of GHL that best represents a group of pedons to be aggregated.

2. Assign GHL to each horizon using whatever information is available for grouping horizons.

3. Evaluate GHL assignments and manually refine as needed.

4. Keep track of final GHL assignments in NASIS or text file.

5. Estimate a most likely horizonation e.g., top and bottom depths for each generalized horizon label.

6. Compute range in characteristics, aka low-rv-high values, for clay, sand, pH, etc. over GHL. (next document in this series)

→ extended tutorial

• a collection of pedons for a component should represent an aggregated component
• a GHL process of some kind is required to efficiently summarize and aggregate horizon data from these collections
• low-rv-high values must be tied to some metric:
  • 5th-50th-95th percentile is one option

• linked pedons represent examples of the aggregate data, but aggregate data should not be limited to a very small set of linked pedons

• IMPORTANT these reports and data summaries are a useful starting point for the aggregation process

• suggested basis for RIC:
  • 5th, 50th, and 95th percentiles are low-rv-high values
  • no assumptions of distribution, simple interpretation

• suggested basis for RIC:
  • 5th, 50th, and 95th percentiles are low-rv-high values
  • no assumptions of distribution, simple interpretation

• 15 pedons correlated to the Loafercreek soil series
• Fine-loamy, mixed, superactive, thermic Ultic Haploxeralfs
• Common soil formed on meta-volcanic rocks of the Sierra Nevada Foothills, MLRA 18
• Included in the soilDB package for testing purposes

• sort by frequency

• sort alphabetically

• plot ranges in horizon depths

• pattern matching via regular expression (REGEX)

  • this is where most micro-correlation decisions are defined

• GHL and rules for our sample dataset:

  • A: ^A$|Ad|Ap
  • Bt1: Bt1$
  • Bt2: ^Bt2$
  • Bt3: ^Bt3|^Bt4|CBt$|BCt$|2Bt|2CB$|^C$
  • Cr: Cr
  • R: R

• special characters in REGEX rules:

  • | = “or”
  • ^ = anchor to left-side
  • $ = anchor to right-side

• cross-tabulation of original names (columns) vs. GHL (rows)
• be sure to check the not-used row

• plot profiles with horizons colored according to GHL
• does it make sense?
• what about horizons in the not-used group?

• plot range in GHL depths, check for overlap

• multivariate summary of depth, clay content, and RF volume

• generalized horizon labels (so far) are stored in R session
• Use an R script to create a text file, called horizon_agg.txt, of horizon ID's and corresponding GHL assignments
• run the NASIS calculation to update the comp layer id field with the GHL for each pedon horizon
• Calculations/Validations –> Pedon Horizon
  • 'Update horizon group aggregations using a text file'
  • Calculation will look for the horizon_agg.txt file in: C:/data/horizon_agg.txt
  • Calculation will load the file and update the comp layer id field
• manual adjustments in NASIS: outliers, special cases, etc.

→ full instructions here

• Stephen's Examples:

  • Lecyr Pedon Report
  • Genesee Lab Data Report
  • Cincinanti Map Unit Report

• Dylan's Examples:

  • Loafercreek Pedon Report
  • Dunstone Pedon Report
  • Amador Pedon Report

Questions, comments, ideas

R resources for pedologists

• → NCSS Job-Aids
• → aqp tutorials
• → Statistical data analysis for pedologists
• → Dylan Beaudette's blog
• → soil-pit Github repository
  
  

Additional AQP Contributors:

• Pierre Roudier (Landcare Research)