Performance of Landsat 8 and Sentinel 2A in vegetation cover mapping of Ise Forest Reserve, Southwest Nigeria

Olaniyi, O.E.1, Adegbola, O.O.1 and Adefurin, O.M.2

1 Department of Ecotourism and Wildlife Management, Federal University of Technology, Akure, Nigeria.

2 Tennessee State University, 330 10th Avenue North, Nashville, TN 37203, United States of America.

Corresponding author: oeolaniyi@futa.edu.ng

Acceptedon June 29, 2020

 

Abstract

In recent years, the high cost and non-affordability of high-resolution satellite imageries had caused high reliability on the medium resolution satellite imageries (MRSI) in developing countries such as Nigeria. Also, the Normalized Difference Vegetation Index (NDVI) had been the most commonly used vegetation index for Vegetation Cover Mapping (VCM). It is pertinent to determine the most accurate vegetation index for VCM of a degraded forest environment in Nigeria despite the varied spatial and spectral resolutions of the MRSI. This study determined the appropriate vegetation indices from Landsat 8 and Sentinel 2A of degraded tropical forest in Ise Forest Reserve, Southwest Nigeria. Recent Landsat 8 and Sentinel 2A satellite imageries were acquired and pre-processed. The earlier was downscaled to ensure uniformity in spatial resolution with Sentinel 2A. Seven vegetation indices: NDVI, Enhanced Vegetation Index (EVI), Green Normalized Difference Vegetation Index (GNDVI), Pigment Specific Simple Ratio (PSSR), Soil Adjusted Vegetative Index (SAVI), Modified Soil Adjusted Vegetative Index (MSAVI) and Transformed Soil Adjusted Vegetative Index (TSAVI), were extracted from the datasets. Both Landsat 8 and Sentinel 2A have the same overall accuracy (98.00%) and kappa coefficients (0.96) in NDVI, SAVI and transformed SAVI. However, EVI (97.69% and 0.854), followed by GNDVI (91.50% and 0.802) extracted from Sentinel 2A outperformed NDVI (a common and widely used vegetation index) based on their overall accuracy and kappa coefficients respectively. The high performance of EVI (0.06 – 0.32) derived from Sentinel 2A despite the downscaling of Landsat 8 makes the consideration of spectral surface reflectance consequential. The enhanced capability and performance of Sentinel 2A in vegetation cover mapping should be more explored in developing economies with low affordability of commercial geospatial data.

 

Keywords: Nigeria, Vegetation indices, Sentinel 2A, Landsat 8, Ise Forest Reserve

 

Download article here