3D modeling reveals how deep drones can “see” into maize canopies
image: Framework of the proposed probe depth method from spectral reflectance.
Credit: The authors
They found that sensing depth shifts across growth stages, with middle leaves contributing most to reflectance and near-infrared signals most sensitive to planting density. The team also developed a hybrid model to estimate vertical LAI distribution, improving multilayer canopy monitoring for precision agriculture.
LAI, the total leaf area per unit ground area, is a core indicator of crop growth, photosynthetic capacity, and yield potential. In maize, leaves form a vertically layered and highly heterogeneous canopy, with different layers contributing unequally to photosynthesis and resource use. Traditional LAI measurements are labor-intensive and limited in scale, while remote sensing provides fast, non-destructive, large-area observations. However, most current approaches treat the canopy as a simple, one-dimensional layer. They often ignore vertical heterogeneity and suffer from problems such as vegetation index saturation in dense crops. For tall, structurally complex species like maize, it remains unclear how deep optical sensors can “see” into the canopy, and how canopy structure and planting density distort the LAI signal detected from above. Addressing these challenges requires 3D modeling of radiation transport and careful linkage between canopy structure and spectral responses.
A study (DOI: 10.1016/j.plaphe.2025.100100) published in Plant Phenomics on 7 September 2025 by Xinming Ma’s team, Henan Agricultural University, offers a practical route to more accurate LAI inversion in real-world, heterogeneous fields, supporting precision nitrogen management, yield forecasting, and stress diagnosis.
Using field measurements, sequential leaf removal, radiative transfer simulations with the LESS model, vegetation index analysis, and the FuseBell-Hybrid inversion model, the study quantified how maize canopy structure influences spectral response and sensing depth. Flat and compact genotypes were evaluated under different planting densities, and leaves were progressively removed from the bottom to determine their individual contributions. LESS simulations were used to track spectral penetration across growth stages, while ten vegetation indices were tested for sensitivity. The FuseBell-Hybrid model was trained on 650 multi-year samples for vertical LAI estimation. Results showed notable structural differences: under high density, flat maize lost bottom leaf area rapidly (15% lower at the 6-leaf stage and <0.05 m²/m² by R1), whereas compact maize maintained larger middle-layer area (0.4 m²/m² in layers 6–8) and retained >0.3 m²/m² across 80% of canopy depth at R3, yielding 40% more total leaf area. LAI declined from 6.9 to 0.91 during stepwise leaf removal, with strong sensitivity when mid-layers were removed. Canopy function showed two thresholds—10–11 leaves sustaining high efficiency, while ≤8 leaves shifted reflectance toward soil dominance. Sensing depth ranged from the lowest 1–3 leaves at seedling stage to 1–7 leaves during grain filling, with middle leaves (5–12) most influential. MCARI was highly density-responsive, MTVI2 performed best for middle-layer monitoring, and NDVI showed saturation in lower layers. The FuseBell-Hybrid model achieved strong prediction accuracy (R² ≈ 0.71–0.77), outperforming traditional VI-based methods and demonstrating reliable capacity for heterogeneous canopy inversion despite decreased accuracy in 2022 due to senescence effects.
These findings provide a structural “map” of how maize canopies are sensed by UAV-borne spectral cameras. By clarifying which leaf layers dominate the spectral signal at different stages and densities, the study helps agronomists and remote sensing practitioners design better monitoring strategies—choosing spectral bands and vegetation indices that are most informative for specific canopy layers. The results emphasize the value of near-infrared bands and structurally sensitive indices like MCARI and MTVI2 for multilayer monitoring, while highlighting the limitations of NDVI in dense maize stands.
###
References
DOI
Original Source URl
https://doi.org/10.1016/j.plaphe.2025.100100
Funding information
This work was partially supported by the Natural Science Foundation of China (32401705), the China Postdoctoral Science Foundation (GZC20230721), the National Key Research and Development Program of China (2024YFD2301100), and the Special Project of Key Research and Development Program of Henan Province (2511111000).
About Plant Phenomics
Plant Phenomics is dedicated to publishing novel research that will advance all aspects of plant phenotyping from the cell to the plant population levels using innovative combinations of sensor systems and data analytics. Plant Phenomics aims also to connect phenomics to other science domains, such as genomics, genetics, physiology, molecular biology, bioinformatics, statistics, mathematics, and computer sciences. Plant Phenomics should thus contribute to advance plant sciences and agriculture/forestry/horticulture by addressing key scientific challenges in the area of plant phenomics.