Calibrated PATH-Based Modeling and Overlapping Cylindrical Voxelization for Maize LAI Mapping Using UAS-LiDAR

ORCID

Yadav: https://orcid.org/0000-0002-8158-7207; Wijewardane: https://orcid.org/0000-0001-8962-9451; Zhang: https://orcid.org/0000-0001-9654-3859; McCraine: https://orcid.org/0000-0002-5849-3616; Silva: https://orcid.org/0009-0006-4333-5128; Dhillon: https://orcid.org/0000-0002-6260-5174

MSU Affiliation

College of Agriculture and Life Sciences; Department of Agricultural and Biological Engineering; James Worth Bagley College of Engineering; Geosystems Research Institute; Department of Plant and Soil Sciences

Creation Date

2026-06-30

Abstract

Accurate retrieval of leaf area index (LAI) in row crops using uncrewed aerial system-light detection and ranging (UAS-LiDAR) requires canopy sampling strategies and radiative transfer formulations that are robust to structural heterogeneity and sparse sampling. This study proposes an integrated framework for high-resolution LAI estimation in maize that combines an adaptive kernel density estimate for localized ground–vegetation separation, overlapping cylindrical voxelization for spatially stable canopy sampling, and a calibrated path-length distribution (PATH) model to explicitly address within- and between-canopy clumping. The voxelization strategy was evaluated against conventional nonoverlapping cubical voxelization using spatial completeness and structural fidelity metrics. For identical LiDAR subsets and grid spacing, overlapping cylindrical footprints achieved higher spatial completeness (fill ratio = 0.36 versus 0.28) and improved agreement with reference imagery (spatial similarity index = 0.64 versus 0.55), while increasing point aggregation without degrading vertical fidelity. Sensitivity analyses using three idealized canopy geometries (cylindrical, conical, and spherical) across multiple nitrogen treatments and growth stages demonstrate that canopy geometry strongly governs LAI bias through path-length variability and foliage clumping. Cylindrical geometry, representing a limiting case with constant path length under near-nadir viewing, exhibited negligible within-canopy clumping bias and high numerical stability across canopy densities. In contrast, conical and spherical geometries showed increased sensitivity to clumping, resulting in systematic LAI underestimation, with total bias reaching approximately 12.7% under dense and high-nitrogen conditions. Calibration of PATH-derived leaf area density resolved scale ambiguity, requiring no correction for cylindrical geometry but substantial geometry-dependent scaling for conical and spherical cases. Validation against in situ measurements confirmed that the cylindrical formulation produced the most reliable LAI estimates (R = 0.85; RMSE = 0.36 m2/m2), supporting its suitability for operational precision-agriculture applications.

Publication Date

4-29-2026

Publication Title

IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing

Publisher

Institute of Electrical and Electronics Engineers

First Page

17894

Last Page

17915

Creative Commons License

Creative Commons Attribution 4.0 International License
This work is licensed under a Creative Commons Attribution 4.0 International License.

Rights

© 2026 The Authors

Link to Full Text

Share

COinS
 

Digital Object Identifier (DOI)

https://doi.org/10.1109/JSTARS.2026.3688894