Mapping land-use pattern using image processing techniques for Medium resolution satellite data: Case study in Matara District, Sri Lanka

Last few decades the land-use pattern has changed in Matara District in Sri Lanka due to the influence of both natural and anthropogenic hazard impacts. These hazards have caused the interruption for social, economic and the environmental sectors in this area. Since the comprehensive field surveys are highly expensive, time consuming and requiring more human power the remote sensing satellite images can be used to map the changing pattern of land-use types over a large period of time. This study focuses on how far remote sensing data give the information on land-use pattern change as an effective information and communication technology in collaboration with GIS. The Medium resolution Lansdsat satellite images were used in this study because they are freely and readily available therein cost effective and highly efficient. The ground features have been extracted from the images in 1990, 1993, 2001, 2003, 2005, 2009 and 2014 respectively. Water body has been extracted using NIR band by using single band classification method while the vegetation cover was extracted by using Normalized Difference Vegetation Index (NDVI). Multiband image classification was also implemented for each year using Maximum Likelihood Classification. According to the results, the land-use pattern has changed over the last 24 years in a significant level. The accuracy of the Maximum Likelihood Classification is 84.43%.