W.50 Dataset

Description

  • Target Soil Properties: SOC, pH, Clay

  • Groups of Features: DEM, ERa, VI, XRF

  • Sample size: 50

  • Number of Features: 15

  • Coordinates: Without coordinates because of privacy concerns

  • Location: Wisconsin, USA

  • Sampling Design: Conditioned latin hypercube sampling based on electrical conductivity, terrain parameters, and normalized difference vegetation index

  • Study Area Size: 80 ha

  • Geological Setting: Glacial outwash and sediments of the Johnson End Moraine

  • Previous Data Publication: None

  • Contact Information:

  • License: CC BY-SA 4.0

  • Publication/Modification Date (d/m/y): 28.02.25, version 1.0

  • Changelog:

    • Version 1.0 (28.02.25): Initial release

Details

Dataset

The dataset contains the following target soil properties and features:

Target Soil Properties:

SOC - Soil Organic Carbon

  • Code: SOC_target

  • Unit: %

  • Protocol: Measured CO₂ release during dry combustion after removing inorganic carbon with an acid (Nelson and Sommers 1996)

  • Sampling Date: July 2019

  • Sampling Depth: 0 – 10 cm

pH

  • Code: pH_target

  • Unit: Unitless

  • Protocol: Measured in water suspension with a glass electrode with a 1:1 liquid:soil gravimetric ratio (Burt 2014)

  • Sampling Date: July 2019

  • Sampling Depth: 0 – 10 cm

Clay

  • Code: Clay_target

  • Unit: %

  • Protocol: Hydrometer method; separation of the fractions by sieving and sedimentation. Measurement of the separated fractions by weighing the density of the suspension (Gee and Bauder 1979)

  • Sampling Date: July 2019

  • Sampling Depth: 0 – 10 cm

Groups of Features:

DEM – Digital Elevation Model and Terrain Parameters

  • Number of Features: 2

  • Code(s): Altitude, Slope

  • Unit: Altitude in m, Slope in °

  • Sensing: Digital elevation model raster (3 m) based on LiDAR from the “Wisconsin Department of Natural Resources”

  • Processing: Calculating Slope with terrain function of the raster package, extracting DEM values from raster at soil sampling locations, resampled from the original 3 m resolution to 5 m resolution

  • Sampling Date: Unknown

ERa – Apparent Electrical Resistivity

  • Number of Features: 1

  • Code(s): ERa

  • Unit: Ω m

  • Sensing: DUALEM-1HS instrument (DUALEM Inc., Milton, Canada) with exploration depth of 0 - 30 cm, in-situ

  • Processing: Ordinary Kriging to align sensing- with soil sampling locations

  • Sampling Date: July 2019

VI - Vegetation Indices

  • Number of Features: 2

  • Code(s): NDVI, GNDVI

  • Unit: Unitless

  • Sensing: Sentinel-2 image during vegetative period (Level-2A) from “Copernicus Open Access Hub”

  • Processing: Calculating NDVI as (B08 - B04) / (B08 + B04) and GNDVI as (B08 - B03) / (B08 + B03), extracting VI values from raster at soil sampling locations

  • Sampling Date: July 2019

XRF – X-ray Fluorescence Derived Elemental Concentrations

  • Number of Features: 10

  • Code(s): XRF_Mg, XRF_Al, XRF_Si, XRF_Ca, XRF_Ti, XRF_Mn, XRF_Fe, XRF_Zn, XRF_Sr, XRF_Zr

  • Unit: ppm (estimated through XRF Geochem not ground truth)

  • Sensing: Delta Premium PXRF spectrometer (Olympus Scientific Solutions Americas Inc., Waltham, USA), on dried and sieved samples (<2 mm) in the laboratory

  • Processing: Compton normalization method to transform full spectra into estimates of elemental concentrations with accompanied software of the sensor (Geochem mode)

  • Sampling Date: July 2019

Examples

from LimeSoDa import load_dataset, split_dataset
from sklearn.linear_model import LinearRegression
from sklearn.metrics import r2_score, mean_squared_error
import numpy as np

# Load and explore the dataset
data = load_dataset("W.50")
dataset = data["Dataset"]
folds = data["Folds"]
coords = data["Coordinates"]  # Note: No coordinates available

# Split into train/test using fold 1
X_train, X_test, y_train, y_test = split_dataset(
    data=data,
    fold=1,
    targets=["pH_target", "SOC_target", "Clay_target"]
)

# Fit model and get predictions
model = LinearRegression()
model.fit(X_train, y_train)
predictions = model.predict(X_test)

# Calculate performance metrics
r2 = r2_score(y_test, predictions)
rmse = np.sqrt(mean_squared_error(y_test, predictions))
print(f"R-squared: {r2:.7f}")
print(f"RMSE: {rmse:.7f}")

References

Burt, R. (Ed.) (2014). Kellogg soil survey laboratory methods manual. United States Department of Agriculture, Natural Resources Conservation Service, National Soil Survey Center, Kellogg Soil Survey Laboratory.

Gee, G. W., & Bauder, J. W. (1979). Particle size analysis by hydrometer: a simplified method for routine textural analysis and a sensitivity test of measurement parameters. Soil Science Society of America Journal, 43(5), 1004-1007.

Nelson, D.W. & Sommers, L.E. (1996) Total Carbon, Organic Carbon, and Organic Matter. In: Sparks, D.L., Page, A.L., Helmke, P.A., Loeppert, R.H., Soltanpour, P.N., Tabatabai, M.A., Johnston, C.T. & Sumner, M.E., Eds., Methods of Soil Analysis. Part 3. Chemical Methods, Soil Science Society of America, Madison, WI, 961-1010.