IJASEIT

International Journal on Advanced Science, Engineering and Information Technology, Vol. 12 (2022) No. 3, pages: 881-891, DOI:10.18517/ijaseit.12.3.12514

Combining Pan-Sharpening and Forest Cover Density Transformation Methods for Vegetation Mapping using Landsat-8 Satellite Imagery

Projo Danoedoro, Diwyacitta Dirda Gupita

Abstract

Forest cover density (FCD) transformation is an 8-bit Landsat imagery-based method for vegetation mapping, which uses a set of indices comprising vegetation, soil, shadow, and thermal components. With the advent of 16-bit Landsat-8 imagery, radiometric correction and pan-sharpening methods could be applied to generate new datasets with different spectral and spatial characteristics. This study combined several methods of pan-sharpening and FCD transformations for mapping vegetation density in Salatiga and Ambarawa region, Indonesia, based on Landsat-8 dataset. The imagery was treated differently to constitute five new datasets, i.e., original multispectral imagery (30 m), radiometrically corrected multispectral imagery (30 m), and three pan-sharpening datasets generated using Gram-Schmidt (GS), Principal Component Analysis (PCA), and Hyper-spherical Color Space (HCS) methods (15 m). Each dataset was then processed using FCD transformation as a basis for vegetation density and structural composition mapping. Field observation and vegetation density measurement using high-spatial-resolution imagery was used as a reference for accuracy assessment. This study found that the pan-sharpening methods produced new datasets with various correlation coefficients with their corresponding original bands, affecting the accuracy of spectral modeling in FCD. Moreover, the generated FCD models were found less accurate as compared to that of the original one. However, the accuracy could be increased by rescaling the original DNs and regrouping the original classes into simpler categorization. Besides the problem of data characteristics, all FCD models were also found inaccurate compared to previous studies due to the landscape complexity of the study area.