Chiozza et al. (2026) Tracking temporal variations in the soil-plant-atmosphere continuum in wheat using multisensor data

Identification

• Journal: Smart Agricultural Technology
• Year: 2026
• Date: 2026-01-10
• Authors: M. Chiozza, L. Sánchez-Fernández, M. Pérez-Ruiz, G. Egea
• DOI: 10.1016/j.atech.2026.101794

Research Groups

• Universidad de Sevilla, Área de Ingeniería Agroforestal, Dpto. de Ingeniería Aeroespacial y Mecánica de Fluidos, Spain

Short Summary

This study developed an integrated multi-sensor high-throughput phenotyping platform and statistical modeling framework to monitor season-long temporal dynamics of wheat photosynthetic traits and water status, demonstrating high accuracy in predicting net photosynthetic rate and stomatal conductance from hyperspectral data.

Objective

• To design and implement a high-throughput phenotyping platform (HTPP) for real-time, multi-sensor monitoring of the soil-plant-atmosphere continuum (SPAC) under field conditions.
• To establish quantitative associations between sensor-derived data and key wheat physiological traits related to photosynthesis and water status.
• To reconstruct the temporal dynamics of the SPAC from multi-sensor data to support genotype evaluation, cultivar selection, and management optimization in precision agriculture.

Study Configuration

• Spatial Scale: Field experiment at the Campus of the University of Seville, Spain (37.351174, −5.938766). Experimental plots measured 1.3 meters in width by 6 meters in length, with measurements taken over a 4-meter section for hyperspectral data and soil moisture at 0.057 meters depth. The trial included 2 water regimes (full irrigation, rainfed), 4 commercial wheat cultivars, and 3 replications, totaling 24 plots.
• Temporal Scale: The growing season from February 1st, 2024, to May 14th. Crop physiological measurements (net photosynthetic rate, stomatal conductance, vapor pressure deficit) and hyperspectral data were collected on 7 dates (43, 50, 67, 74, 83, 92, and 102 days after sowing (DAS)). Soil moisture and canopy temperature were recorded on 3 dates (74, 83, and 92 DAS). Phenological observations were conducted at 12 intervals throughout the season.

Methodology and Data

• Models used:
  • Partial Least Squares Regression (PLSR) for predicting physiological traits from hyperspectral data.
  • Generalized Additive Models (GAMs) for characterizing temporal dynamics of soil and crop variables.
  • One-way ANOVA for statistical comparisons.
• Data sources:
  • Multi-sensor High-Throughput Phenotyping Platform (HTPP):
    • Hyperspectral camera (Nano-Hyperspec sensor, Headwall Photonics): Reflectance data (397–1003 nm, 197 bands).
    • Automatic soil moisture sensing system (HydraProbe®, Stevens Water Monitoring Systems Inc.): Volumetric soil water content (SVW) at 0.057 meters depth.
    • Infrared thermometry system (MLX90614 infrared sensor, Melexis Technologies): Crop canopy temperature, wet and dry reference surface temperatures for Crop Water Stress Index (CWSI).
    • RTK-GNSS system for centimeter-level positioning and LiDAR.
  • Ground-truth measurements:
    • Portable gas exchange system (CIRAS-3, PP Systems): Leaf net photosynthetic rate (A, µmol CO₂ m⁻² s⁻¹), stomatal conductance (Gs, mmol H₂O m⁻² s⁻¹), and vapor pressure deficit (VPD, kPa).
    • BBCH scale for crop phenology.
  • Derived data: 20 Vegetation Indices (VIs) from hyperspectral data and Crop Water Stress Index (CWSI) from temperature data.

Main Results

• Water treatments effectively established contrasting moisture environments, with irrigated plots showing increased soil moisture and consistently low Crop Water Stress Index (CWSI), while rainfed plots exhibited declining soil moisture and increasing CWSI over time.
• Hyperspectral data accurately predicted net photosynthetic rate (A) and stomatal conductance (Gs) using PLSR models, achieving coefficients of determination (R²) of 0.72 for A and 0.70 for Gs.
• The Root Mean Square Error of Prediction (RMSEP) for A was 3.71 µmol CO₂ m⁻² s⁻¹ and for Gs was 58.92 mmol H₂O m⁻² s⁻¹ kPa⁻¹.
• Generalized Additive Models (GAMs) successfully captured the temporal dynamics of observed and predicted A, Gs, soil moisture, and CWSI, with predicted trends closely matching ground-truth measurements (GAMs explained 87% of deviance for predicted A and 77% for predicted Gs).
• The most influential predictors for both A and Gs were concentrated in the near-infrared region (757–909 nm), with additional significant contributions from blue (477.21 nm for Gs) and red (660.29, 662.50, 664.71, 669.12 nm for A) bands.
• No significant cultivar x treatment interaction was observed for A, Gs, soil moisture, or CWSI, indicating uniform physiological responses to water stress across the tested genotypes under the specific experimental conditions.

Contributions

• Developed and implemented an integrated multi-sensor high-throughput phenotyping platform (HTPP) for real-time, high-resolution monitoring of the soil-plant-atmosphere continuum (SPAC) under field conditions.
• Established a novel framework combining multi-sensor data with statistical modeling (PLSR, GAMs) to reconstruct season-long, fine-scale temporal dynamics of photosynthetic traits (net photosynthetic rate, stomatal conductance) and water status (volumetric soil water content, crop water stress index).
• Demonstrated high accuracy in predicting dynamic net photosynthetic rate and stomatal conductance from hyperspectral data, enabling the extraction of continuous daily trait values and curve-derived metrics for genotype discrimination and stress resilience breeding.
• Provided scalable, data-driven solutions to support breeding for resilient cultivars and improvements in crop management by integrating predicted data into mechanistic crop models.

Funding

• Project PID2021-125080OB-I00 funded by the Spanish Ministry of Science and Innovation.

Citation

@article{Chiozza2026Tracking,
  author = {Chiozza, M. and Sánchez-Fernández, L. and Pérez-Ruiz, M. and Egea, G.},
  title = {Tracking temporal variations in the soil-plant-atmosphere continuum in wheat using multisensor data},
  journal = {Smart Agricultural Technology},
  year = {2026},
  doi = {10.1016/j.atech.2026.101794},
  url = {https://doi.org/10.1016/j.atech.2026.101794}
}