Comparison of remote sensing approach for mangrove mapping over Penang Island

Mangrove mapping is crucial for the policy maker to have more proper planning of land use of a country or nation while helping to preserve the mangrove area. The unique mangrove ecosystem need to be conserved as mangrove trees has many applications for human being not only in their forestry products such as timber and charcoal but also serve as a strong barrier from the attack of wave, erosion and tsunami to inland area near the seashore or coastal region. The aim of this paper is to compare the accuracy of the mangrove map produced from different remote sensing techniques. Thailand Earth Observation System (THEOS) satellite data of Penang Island with date 29 January 2010 was utilized for the image processing analysis. All the pre-processing, classification, validation and post-classification analysis were done by using Geomatica version 10.3.2 software package. The results obtained show that Artificial Neural Network (ANN) with classification accuracy of 93.5% can increase the overall accuracy by 2.0% as compare to Maximum Likelihood Classification method (91.5%). This study indicates that ANN approach which has the highest accuracy and kappa coefficient is more reliable used for mangrove mapping at generic level.