Optimizing canopy cover evaluation

This article introduces a comprehensive framework that uses LiDAR data and machine learning techniques to predict a range of biophysical variables. The framework is tested explicitly on Canopy Cover. The methodology combines optimized feature engineering, selection, and scalable parallel processing...

Descripción completa

Detalles Bibliográficos
Autores: Ventura Rodríguez, Pau|||0009-0009-6441-4804, Carrillo, Carlos|||0000-0003-3606-7517, Donaire Salvador, Alejandro|||0009-0003-0620-6616, Sánchez, Eric
Tipo de recurso: artículo
Fecha de publicación:2026
País:España
Institución:Universitat Autònoma de Barcelona
Repositorio:Dipòsit Digital de Documents de la UAB
Idioma:inglés
OAI Identifier:oai:dnet:uabarcelona_::3d3ebc8a499af55aeff2722d74406f99
Acceso en línea:https://ddd.uab.cat/record/328275
https://dx.doi.org/urn:doi:10.1016/j.envsoft.2026.106982
Access Level:acceso abierto
Palabra clave:Canopy cover
LiDAR data
Machine learning
Descripción
Sumario:This article introduces a comprehensive framework that uses LiDAR data and machine learning techniques to predict a range of biophysical variables. The framework is tested explicitly on Canopy Cover. The methodology combines optimized feature engineering, selection, and scalable parallel processing to ensure both accuracy and computational efficiency. The study demonstrates the efficacy of AI-CanopyMapper in accurately predicting canopy cover. The full model achieved a mean absolute error (MAE) of 6.47% and an of 0.88. In contrast, the partial model, trained with only 1.3% of the available data, reached a MAE of approximately 15%, demonstrating strong generalization capabilities even under data-limited conditions. These results confirm that AI-CanopyMapper offers a fast, scalable, and data-efficient framework compared to traditional methods, highlighting its potential applications in forest management and environmental monitoring.