Remote sensing technique for estimating the age of oil palm using high resolution image

Oil palm is a tropical crop that mainly grows for its industrial production and has become one of the most economic crops in some countries especially in Malaysia and Indonesia. Although oil palm is one of Malaysia's major sources of revenue, there are lots of controversy about the impact of this massive cultivation on the environment. Age of the Oil Palm is one of the most important factors that plays a significant role in the productivity of the oil palm as well as for the impact assessment of oil palm plantation on carbon sequestration process. Although the age of the oil palm is available to the respective oil palm companies, this information is very hard to obtain collectively from the different oil palm companies for research purposes. Therefore, this study is going to use remote sensing technique for the estimation of the age of the oil palm using a high-resolution satellite data (Worldview-2). The complexity of the determination of oil palm age from the satellite data is that age information cannot be extracted easily due to the similar spectral signature among the different oil palm age groups as well as confusion with the spectral signature of forest and other features. Various image processing techniques such as band rationing, vegetation indices and texture measurement were carried out for generating variable image parameters from the original data in order to overcome the spectral confusion among the different age groups as well as with other features. All the image parameters were classified using Maximum Likelihood Classifier (MLC) and Artificial Neural Network (ANN). The results of the classification were validated using the ground truth data, and comparison of different results was made to find the best technique that can be used for the determination of the oil palm age effectively. The result indicates that multispectral band is helpful to determine the age of the oil palm, but obtained accuracy was not promising because of the similar spectral signature among the different age classes, however, the classification accuracy was improved when the textural information was used in the classification algorithm. This research found that texture measurement is very promising, and it has the potential to differentiate different features as well as the age group of oil palm plantation, and an accuracy of 60% was obtained using multi-scale and multi-texture approach