ASSESSING THE APPLICABILITY OF GLOBAL FOREST COVER CHANGE DATASETS TO NATIONAL FOREST MONITORING: THE CASE OF ETHIOPIA

Abstract

Accurate and up-to-date global land cover change datasets are most important in monitoring forest cover land use changes. Such data sets are particularly important for developing countries with huge data gaps. In this study, two existing global forest cover land use datasets derived from FAO Remote Sensing Survey (RSS) and Global Forest Cover Change (GFCC), with the same Landsat sensor data sources and 30-meter spatial resolution were compared to evaluate their accuracy and their adaptability to national forest monitoring systems (NFMS) of Ethiopia. The study employed systematic sampling to get access to available RSS tiles and simple random sampling to incorporate GFCC datasets outside the selected tiles. The results of the study revealed that forest definitions and methodologies for each dataset employed vary, underpinning the variation of forest cover extent and forest loss rate for the same period. The accuracy analysis result indicated that 61% of GFCC with Kappa of 0.50 and 80% of RSS datasets with kappa of 0.74 have overall accuracy and agreement with reference data and are reasonably acceptable. The overall spatial agreement between the two datasets was 53%, which is less than the accuracy observed. In conclusion, the findings imply that, although both datasets are consistent in their methodology for adaptability, the RSS dataset is more accurate and comparable, while the GFCC dataset is complete and transparent to be used for measuring, reporting, and verification of national forest monitoring activities primarily for reducing emissions from deforestation and forest degradation at national level in Ethiopia.