Skip to main content
ARS Home » Plains Area » Temple, Texas » Grassland Soil and Water Research Laboratory » Research » Publications at this Location » Publication #418884
Research Project: Enhancing Cropping System and Grassland Sustainability in the Texas Gulf Coast Region by Managing Systems for Productivity and Resilience

Location: Grassland Soil and Water Research Laboratory

Title: Enhancing leaf area index estimation with multispectral imagery and machine learning

Author
item RAM SAPKOTA, BALA - Texas Agrilife Research
item CHATTERJEE, SUMANTRA - Texas Agrilife Research
item BAATH, GURJINDER - Texas Agrilife Research
item Flynn, Kyle
item Smith, Douglas

Submitted to: Meeting Abstract
Publication Type: Abstract Only
Publication Acceptance Date: 8/29/2024
Publication Date: N/A
Citation: N/A

Interpretive Summary: n/a - abstract only.

Technical Abstract: Leaf Area index (LAI) is a critical growth parameter in precision agriculture , but measuring it in large fields and multiple locations can be difficult. Unmanned Aerial Vehicle (UAS) based multispectral imagery has been used to estimate LAI using vegetative Indices (VIs), but often face issues in dense canopies. This study used diverse LAI data from corn plots across three locations (2022 -2023) to compare reflectance-based and VI-based models using five machine learning (ML) algorithms. Models were trained and tested using the data ratio of 80:20. Results showed reluctance-based models, especially those using red edge and near-infrared bands, outperformed VI-based models at mid and late vegetative stages and early reproductive stages. VI-based models performed well at early vegetative stages. K-Neighbours and Extra Tree regressors were most effective on LAI estimation. Results suggest that using machine learning to improve canopy spectral reflectance improves LAI predictions, which helps with precision agriculture in dense crops like corn.