EVALUATING THE EFFICIENCY OF PLEADDES AND SPOT 6 MULTISPECTRAL FUSION IMAGE FOR MAPPING CORAL REEF SUBSTRATES IN SELINGAN ISLAND, SABAH, MALAYSIA

摘要

Mapping coral reef substrate using traditional ground base method is inefficient and very costly. However, satellite remote sensing is one of the technologies that have the potential to map accurately the distribution of coral reef substrates. The spatial component of the coral reef can be delineated using high resolution satellite imagery. Coral reef diversity has rapidly degradating over the last decade so the need for an accurate distribution coral reef substrates map is very crucial for better planning, management and monitoring of coral reef area. This paper evaluated the efficiency of using image fusion techniques in the mapping of coral reef substrates. The aim of the study was to determine and compare the accuracy of coral reef substrates classification between multispectral and fusion image of Pleaides and SPOT 6 data. Multispectral data of high resolution Pleaides and SPOT 6 satellite imageries were used in the study. Several radiometric correction techniques such as conversion from digital number to radiance and reflectance, atmospheric correction, water column correction and sun glint correction were applied to the individual Pleaides and SPOT 6 multispectral images and fusion images. The enhanced imageries were then classified using maximum likelihood classification in order to generate coral reef substrates map. The accuracy of each coral reef substrates classification map was subsequently derived from the accuracy assessment for each classification. The study had found that Pleaides data produced the most accurate results of coral reef substrates classification with 89.20% of overall accuracy. This was followed by fusion image of Pleaides with SPOT 6 which produced 87.20% of overall accuracy. SPOT 6 data produced the lowest accuracy of coral reef substrates classification with 86.80% of overall accuracy. The maximum likelihood classification clearly distinguished coral reef substrates within the study area into five classes, namely, live coral, dead coral, rubble, sand and unclassified. The study concludes that the fused multispectral image of Pleaides with SPOT 6 did not improved the accuracy level of coral reef classification compared with the original image of Pleaides which produced the highest overall accuracy.