Linear spectral mixture analysis of land cover for assessment level sub-pixel using PlanetScope imagery in Dieng Plateau

Abstract

Abstract Information on the land cover is essential as thematic information for planning, controlling, and spatial planning to create sustainable development. Remote sensing is a tool used to solve a problem related to space, environment, and zoning. One of the obstacles in using remote sensing data is the presence of mixed pixels. This research was conducted in the Dieng Plateau. This research aims to calculate accuracy assessment and land cover map fraction using PlanetScope imagery in Dieng Plateau. The remote sensing image is PlanetScope, recorded November 19, 2020. Topographic correction is carried out using the cosine correction method. The Linear Spectral Mixture Analysis (LSMA) method determines each endmember percentage land cover fraction. The field survey was conducted with 74 samples for accuracy testing. Each sample plot measures 10 m x 10 m. The topographic correction results can significantly affect land cover classification results. The result of the LSMA method classification used PlanetScope images to determine the percentage land cover map fraction of bare soil, impervious surface, vegetation, and water in one pixel with an accuracy of MAE 99.76%.