Validation-Aware VNIR Spectroscopy Reveals Target-Specific Transferability of SOC, SIC, and Gypsum Predictions in Dryland Agricultural Soils
Abstract
Hyper-arid agricultural soils combine low organic carbon, high calcium carbonate contents, salinity, and frequent gypsum accumulation, making rapid quantification of soil organic carbon (SOC), soil inorganic carbon (SIC), and gypsum valuable but analytically demanding. We evaluated visible–near infrared (VNIR) spectroscopy for predicting these three properties in 216 samples from 18 sites spanning two oasis management systems, three topographic positions, and six soil depths, with particular emphasis on how validation design affects apparent model accuracy. Partial least squares regression (PLSR) and random forest (RF), single- and multi-target workflows, and global versus stratified calibrations were compared using site-based GroupKFold, random KFold, and leave-one-out cross-validation (LOOCV). RF outperformed PLSR across all targets. Predictability was strongly target-specific: gypsum was predicted most reliably (R² = 0.79), SOC moderately (R² = 0.46), and SIC least reliably (R² = 0.36), with stratified SIC models often failing under site-blocked validation. KFold and LOOCV consistently yielded more optimistic estimates than GroupKFold, inflating R² by up to 0.17 for SIC. VNIR spectroscopy therefore shows strong potential for gypsum, moderate utility for SOC, but limited transferability for SIC. Validation design critically determines how soil-spectroscopy performance should be interpreted for dryland monitoring.
Related
- Multi-Task Graph Neural Network Predictions of Auger-Electron and X-ray Photoelectron Spectroscopy
- Donor-level validation of Raman spectroscopy for breast cancer subtype classification
- Simultaneous Site- and Residue-Specific Incorporation of Fluorinated Amino Acids for In-Cell Structural Biology by 19F Magic Angle Spinning NMR Spectroscopy
- Site-Specific Speciation of Vanadium Sites on γ-Al2O3 Facets and Edges from 51V SSNMR Spectroscopy
- Target-Specific De Novo Drug Design via Fine-Tuned Language Models and Molecular Docking, Molecular Dynamics Simulation Validation
BibTeX
@misc{allagui2026validationaware,
title = {{Validation-Aware VNIR Spectroscopy Reveals Target-Specific Transferability of SOC, SIC, and Gypsum Predictions in Dryland Agricultural Soils}},
author = {Wajdi Allagui and Hayfa Zayani and Luca Bernardini and Pascal Pichelin and Nadhem Brahim and Roland Bol and Hamouda Aichi and Youssef Fouad and Wolfgang Wanek},
howpublished = {HAL},
year = {2026},
month = sep,
doi = {10.20944/preprints202609.0396.v1},
url = {https://institut-agro-rennes-angers.hal.science/hal-05741945v1},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Validation-Aware VNIR Spectroscopy Reveals Target-Specific Transferability of SOC, SIC, and Gypsum Predictions in Dryland Agricultural Soils
AU - Wajdi Allagui
AU - Hayfa Zayani
AU - Luca Bernardini
AU - Pascal Pichelin
AU - Nadhem Brahim
AU - Roland Bol
AU - Hamouda Aichi
AU - Youssef Fouad
AU - Wolfgang Wanek
PY - 2026
DA - 2026/09/04
PB - HAL
DO - 10.20944/preprints202609.0396.v1
UR - https://institut-agro-rennes-angers.hal.science/hal-05741945v1
AB - Hyper-arid agricultural soils combine low organic carbon, high calcium carbonate contents, salinity, and frequent gypsum accumulation, making rapid quantification of soil organic carbon (SOC), soil inorganic carbon (SIC), and gypsum valuable but analytically demanding. We evaluated visible–near infrared (VNIR) spectroscopy for predicting these three properties in 216 samples from 18 sites spanning two oasis management systems, three topographic positions, and six soil depths, with particular emphasis on how validation design affects apparent model accuracy. Partial least squares regression (PLSR) and random forest (RF), single- and multi-target workflows, and global versus stratified calibrations were compared using site-based GroupKFold, random KFold, and leave-one-out cross-validation (LOOCV). RF outperformed PLSR across all targets. Predictability was strongly target-specific: gypsum was predicted most reliably (R² = 0.79), SOC moderately (R² = 0.46), and SIC least reliably (R² = 0.36), with stratified SIC models often failing under site-blocked validation. KFold and LOOCV consistently yielded more optimistic estimates than GroupKFold, inflating R² by up to 0.17 for SIC. VNIR spectroscopy therefore shows strong potential for gypsum, moderate utility for SOC, but limited transferability for SIC. Validation design critically determines how soil-spectroscopy performance should be interpreted for dryland monitoring.
ER -