Evaluation Methods of Change Detection of Seagrass Beds in the Waters of Pajenekang and Gusung Selayar

Abstract

Knowledge about coastal and small island ecosystems is increasing for the monitoring of marine resources based on remote sensing. Remote sensing data provides up-to-date information with various resolutions when detecting changes in ecosystems. Studies have defined a shift in marine resources but were limited only to pixel or object classification in changes of seagrass area. In the present study, two classification method analysis approaches were compared to obtain optimum results in detecting changes in seagrass extent. It aimed to determine the dynamics of a seagrass ecosystem by comparing two classification methods in the waters of Gusung Island and Pajenekang, South Sulawesi, these methods being pixel-based and object-based classification methods. This research used SPOT-7 satellite imagery with 6 m2 of spatial resolution. Accuracy assessment using the confusion matrix showed optimum accuracy in object-based classification with an accuracy value of 87 %. Meanwhile, pixel-based classification showed an accuracy value of 78 % around Gusung Island. Pajenekang Island had accuracy values of 69 % with object-based classification and 65 % with pixel-based classification. A comparison of both classification methods revealed statistically high accuracy in mapping the benthic habitats of seagrass ecosystems. The results of the classifications showed a decline in the area of seagrass populations around Gusung Island from 2016 - 2018 and around Pajenekang Island from 2013 - 2017, with a change rate of 11.8 % around the island of Gusung and 7.6 % around the island of Pajenekang. This can explain the reason for the temporal method of object-based research classification having the best potential to process data changes in areas of seagrass in South Sulawesi waters and remote sensing information for the mapping of coastal area ecosystems.
 HIGHLIGHTS
 
 Information on coastal ecosystems globally with remote sensing data is currently very easy to access, but information related to ecosystem management and seagrass ecology in certain areas is still limited
 Analysis of seagrass benthic changes in shallow water requires data processing methods with high accuracy
 The OBIA (Object Based Image Analysis) method is one of the analytical methods that can provide optimal results in observing changes in seagrass ecosystems in the waters of South Sulawesi, Indonesia
 
 GRAPHICAL ABSTRACT