Application of Parametric and Non Parametric Classifiers for Assessing Land Use/Land Cover Categories in Cocoa Landscape of Juaboso and Bia West Districts of Ghana

Satellite image classification has been used for long time in the field of remote sensing since results are environmental research, agriculture, climate change and natural resource management. The cocoa landscape Ghana is complex diverse nature, composing mixture closed forest, open settlements, croplands farms which make mapping difficult. purpose this research to assess compare performances three machine learning classifiers: Support Vector Machine (SVM), Random Forest (RF), Artificial Neural Network (ANN) a statistical algorithm: Maximum Likelihood (ML) know classifier best suited using Juaboso Bia West districts as study area. A representative sampling approach was adopted collect 1246 sample points various Land Use/Land Cover (LULC) types. These were divided at random into 869 form 70% 377 constitute 30% total validation. Stacked sentinel-2 image, data validation storing identities LULC classes imported R run supervised each classifier. show that highest overall accuracy kappa statistics produced by support vector (86.47%, 0.7902); next artificial neural network (85.15%, 0.7700), followed forest (84.08%, 0.7559) finally maximum likelihood (78.51%, 0.6668). final map under can be monitor driven deforestation especially gazetted game reserves. This will also very useful national monitoring framework REDD + project.