Evaluation of hyperspectral remote sensing relevance to estimate oil palm trees nutrition status remote sensing

Lelong Camille, Lanore Mathieu, Caliman Jean-Pierre. 2006. Evaluation of hyperspectral remote sensing relevance to estimate oil palm trees nutrition status remote sensing. In : Second recent advances in quantitative remote sensing (RAQRS'II), Auditori de Torrent, Spain, 25-29 September 2006. Sobrino José A. (ed.). Universidad de Valencia. Valence : Universitat de Valencia, 147-152. ISBN 978-84-370-6533-5 International Symposium on Recent Advances in Quantitative Remote Sensing. 2, Valence, Espagne, 25 Septembre 2006/29 Septembre 2006.

Résumé : This studies focuses on the relationships between the reflectance spectra of oil palm leaves and their deficiencies in nitrogen and different minerals (P, K, Mg, Fe). The aim of this work is to develop tools for nutritive stress detection based on remote sensing. A data base was constituted in Indonesia on oil palm trees showing apparent deficiencies and on others grown in trials balancing N and P on one side and K and Mg on the other side. Measurements were done at the leaflet level, providing with a mean reflectance for the leaf in the visible and near-infrared domain (400-900 nm). In parallel, chemical analysis of the leaflets were achieved to provide with the mean concentrations of the different constituents of the leaf. 48 spectral indexes were selected to describe the more exhaustively the spectral features, or found in the literature as efficient parameters to detect nutrition stresses. Statistical analysis was achieved in two different ways: one to establish a predictive model for the chemical concentrations of N, P, K, Mg and Fe in the leaf and one to discriminate these five main classes of deficiencies observed in the fields. None of these approaches led to a significant result, as errors are very high and much above the stress detection threshold or the expected level of discrimination. Possible causes of noise are analysed and perspective to improve the analysis are given.

Mots-clés Agrovoc : Elaeis guineensis, télédétection, stress, nutrition des plantes

Classification Agris : U30 - Méthodes de recherche
F61 - Physiologie végétale - Nutrition

Auteurs et affiliations

• Lelong Camille, CIRAD-AMIS-UPR Spatialisation (FRA) ORCID: 0000-0002-4850-1010
• Lanore Mathieu, CIRAD-AMIS-UPR Spatialisation (FRA)
• Caliman Jean-Pierre, CIRAD-CP-UPR Systèmes de pérennes (IDN)

Source : Cirad - Agritrop (https://agritrop.cirad.fr/539313/)

Métriques

[ Page générée et mise en cache le 2025-02-18 ]