A Study of urban expansion with random forest techniques : A case study of Mueang Songkhla District, Songkhla Province

Abstract

The aim of this study used remote sensing technology to study the urban expansion in the Songkhla city municipality and Khao Roop Chang subdistrict municipality. The data urban expansion was collected in three periods: Landsat-5 (1999), Landsat-8 (2013) and Sentinel-2 (2020) by random forest and time series methods, followed by validating data from field survey and google earth.

The results showed that thematic map 1999 was classified as building and build up area, equal to 10.371 sq.km. or 28.21 percent of the total area. Meanwhile, overall accuracy is equal to the 80 percent. While, during to 2013 showed that building and buildup area, equal to 20.039 sq.km. or 54.51 percent of the total area. And, overall accuracy is equal to the 85 percent. In addition, during to 2020 showed that building and buildup area, equal to 20.589 sq.km. or 56.01 percent of the total area. Furthermore, overall accuracy is equal to the 82 percent. A summary of this study found that urban expansion over 16 years of Songkhla city municipality and Khao Roop Chang subdistrict municipality was approximately 10.218 sq.km. or 27.80 percent of the total area.