Coastal Change Detection Monitoring Technics: A Case Study of Klong Hok Village, Ko Proet Subdistrict, Laem Sing District, Chanthaburi Province

Abstract

     The Chanthaburi province has 87 km shoreline that covers the area of Tha Mai, Laem Sing, and Khlung districts. Its abundance of marine resources is valuable to the economy, ecology, and society. Currently, the shoreline has been used to support human activities that lead to effects on environmental quality impact, especially coastal LULC change. Therefore, this study has focus on the remote sensing data analysis and digital image processing technics to monitor the coastal change in Klong Hok Village, Ko Proet Subdistrict, Laem Sing District by using the Landsat images in 1990 and 2004. The objectives of the study are (1) to monitor the change of coastal using combination of maximum likelihood and object-based classification, the digital change detection, and accuracy assessment, (2) to assess the optimum classification technique, and (3) to distribute remote sensing technology and the results of this research to support coastal zone sustainable management include conservation and restoration. The methodology of the study has six main steps, which are 1) defining the study area, 2) data acquisition and data preprocessing, 3) maximum likelihood and object-based classification, 4) digital change detection, 5) ground surveying and accuracy assessment, and 6) optimum classification technique

     The results of the maximum likelihood and object-based classification of 1990 and 2004 datasets illustrated from-to change in LULC, included 25 classes and it composed of five different land uses including water body, aquaculture land, bare land, village, and mangrove forest. The maximum likelihood illustrated from-to change in LULC increase including water body, aquaculture land, urban, and mangrove forest were 1.53, 0.38, 0.19, and 2.30 sq.km, respectively while LULC decrease including bare land was 4.40 sq.km The object-based illustrated from-to change in LULC increase including water body, aquaculture land, urban, and mangrove forest were 1.78, 0.06, 0.59, and 2.35 sq.km, respectively while LULC decrease including bare land was 4.79 sq.km. The results of the accuracy assessment of the maximum likelihood and object-based classification demonstrated that the overall accuracy were 77.63%, 67.86%, and the Kappa coefficient were 84.75% and 77.32%, respectively. Here, the object-based classification was considered as the optimum classification technique for monitor coastal erosion based on a high overall accuracy, kappa coefficient, and Z statistics.