Land use land cover change analysis using minimum distance classification (MCD) and spectral angle mapper (SAM) method

Abstract

Increasing population growth followed by massive land use change. These conditions can lead to disturbances in the hydrological function of a watershed. Several spatial approaches are used to determine the extent of land-use change, including Minimum Distance Classification (MDC) and Spectral Angle Mapper (SAM) algorithms. This study aims to gain insights into the performance of the two algorithms in classifying land use change. The study site was conducted in the Putih Sub-watershed, which is part of the upstream Serayu watershed. Land cover classification is divided into three classes (i.e. vegetation/forest, agriculture, and settlement. Spatial data from Landsat 7 and 8 imagery over 10 years (2009-2019) is applied as material for the analysis. Based on the results of the MDC classification, the area of agricultural land increased between 2009 (63%) and 2014 (70%), but decreased in 2019 (66%). While in SAM the same pattern was observed, namely an increase in utilized agricultural area from 2009 (63%) to 2014 (75%) before declining by 73% in 2019. The SAM method (RMSE=75.14) is more accurate than the MDC (RMSE=105.13), although both methods share the same pattern. It is possible that the increase in agricultural land and settlements effect to the severe erosion in the Serayu watershed, which requires the attention of policy makers and related stakeholders.