Assessing land use and land cover dynamics using composites of spectral indices and principal component analysis: A case study in middle Awash subbasin, Ethiopia

Abstract

As of recent decades land use and land cover (LULC) change studies have received much attention partly because of the global effect on climate change and mainly because of the local effect on the health and sustainable provision of ecosystem services. Nonrenewable resources are deteriorating from time to time due to the coupled pressure from population growth and global climate change. Monitoring LULC dynamics and setting appropriate management plans have become crucial, which required timely and accurate information. In order to predict future LULC change impacts and simultaneously fulfill daily demand of the growing population, land use planners and policy makers require accurate information on regular basis. Since the last couple of decades application of remote sensing for earth resource management has become popular and a number of earth observation satellites have been developed and launched to the space to collect land resource data of varied temporal and spatial resolution. The knowledge level to interpret these data have also increased and helped to monitor ecosystem structure and its health status at low cost and less effort. In this study we have employed remote sensing and GIS techniques integrating with ground information to analyze LULC dynamics in Middle Awash subbasin from year 1995 to 2017. Eight different spectral indices were algebraically calculated and best composites were used for land feature identification. Principal component analysis (PCA) has also helped to extract the most useful spectral information by compressing redundant data embedded in each spectral channel. The result has greatly assisted to easily identify and trace training polygons for machine learning processes. An overall classification accuracy of 91.8% with kappa coefficient of 0.89 was achieved after applying the aforementioned techniques based on the chosen maximum likelihood supervised image classification algorithm. The change detection analysis has revealed that Middle Awash subbasin has undergone a significant change, which is attributed mainly to anthropogenic activities. Lake Beseka has expanded along with the expansion of irrigation development in the subbasin creating environmental, social and economic impacts. This study has depicted the importance of compositing spectral indices for better classification accuracy. It has also highlighted the current status of the subbasin from which policy makers could extract basic information for policy amendments.