Assessment of health of oil palm plantation from space using rule based classification

Bong, Sea Poh and Abdul Rashid Mohamed Shariff and Radzali Mispan and Shattri Mansor and Noordin Ahmad (2003) Assessment of health of oil palm plantation from space using rule based classification. [Proceedings Paper]

This research deals with providing intelligent information to the oil palm plantation managers about the health of the palm trees in their plantation. For large plantations, it is not practical nor effective to monitor the entire plantation from ground observations only, as there can be delay in identifying problem areas that need remedial care. This research utilizes imageries obtained from imaging satellites as well as a sample of the ground data. These data are then integrated into a Rule Based System that utilizes known domain knowledge about the oil palm plantation. For the study area in Miri, Sarawak, the Rule Based System developed in this project was able to differentiate areas that had good oil palm trees, satisfactory oil palm trees and poor oil palm trees as well as flooded areas and the other categories of land use in the plantation to an accuracy of 87%. A second study area in Johar was tested with this method and it provided an accuracy of classification of 84% with the method developed in this research. Comparison with the conventional methods of Maximum Likelihood (MLH) and the Parallelpiped Classifier showed the Rule Based Method to be at least 10% better. The system developed in this research is a significant one as it will allow better visualization of the palm oil plantation to aid better management which will result in increased and better yields.

Full text available from:

Abstract

This research deals with providing intelligent information to the oil palm plantation managers about the health of the palm trees in their plantation. For large plantations, it is not practical nor effective to monitor the entire plantation from ground observations only, as there can be delay in identifying problem areas that need remedial care. This research utilizes imageries obtained from imaging satellites as well as a sample of the ground data. These data are then integrated into a Rule Based System that utilizes known domain knowledge about the oil palm plantation. For the study area in Miri, Sarawak, the Rule Based System developed in this project was able to differentiate areas that had good oil palm trees, satisfactory oil palm trees and poor oil palm trees as well as flooded areas and the other categories of land use in the plantation to an accuracy of 87%. A second study area in Johar was tested with this method and it provided an accuracy of classification of 84% with the method developed in this research. Comparison with the conventional methods of Maximum Likelihood (MLH) and the Parallelpiped Classifier showed the Rule Based Method to be at least 10% better. The system developed in this research is a significant one as it will allow better visualization of the palm oil plantation to aid better management which will result in increased and better yields.

[error in script]

Item Type:	Proceedings Paper
Additional Information:	Available at Perpustakaan Sultan Abdul Samad, Universiti Putra Malaysia, 43400 UPM Serdang, Selangor, Malaysia. TP684 P3I61 2003 Call Number
AGROVOC Term:	oil palm > oil palm Prefer using Elaeis guineensisElaeis guineensis
AGROVOC Term:	plantations
AGROVOC Term:	remote sensing
AGROVOC Term:	satellite imagery
AGROVOC Term:	monitoring and evaluation
AGROVOC Term:	image analysis
AGROVOC Term:	researchers > researchers Prefer using scientistsscientists
AGROVOC Term:	scientists
AGROVOC Term:	forest health management
Geographical Term:	Malaysia
Depositing User:	Nor Hasnita Abdul Samat
Date Deposited:	04 Aug 2024 10:08
Last Modified:	04 Aug 2024 10:08
URI:	http://webagris.upm.edu.my/id/eprint/888

Actions (login required)

View Item