Estimation and mapping of fossil fuel carbon emissions using night time light and population data in Uganda

Abstract

Scientific research on fossil fuel carbon emission has emerged as a high priority in climate studies, bio-geochemistry, health studies, environmental assessment and global change policy. However, the estimation and mapping of fossil fuel carbon emissions is still a challenging task in Uganda whose population and economic activities are growing in a fast, almost poorly planned and unregulated manner. The major aim of this study was to explore the potential of using Night Time Lights provided by DMSP/OLS and population provided by NOAA as proxy measures for estimating and mapping fossil fuel carbon emissions in Uganda. Geometric correction and intercalibration were done to improve continuity and comparability of the NTLs data. The Ghosh et al. (2010) model which allocates spatial emission sources from a large geographic area to finer grid cells was used to disaggregate the emissions. The lit pixels were multiplied by emission per radiance and the unlit pixels were multiplied by the emission per capita. The two datasets were merged in ArcGIS software to obtain the total CO2 emission maps. The distributions of the emissions were established with the highest obtained in the built up areas, farm lands, open waters and the lowest found in the tree and forest plantations. The trend of fossil fuel carbon emission (1992-2013) was at R2 = 0.906. A very strong relationship of R2 = 0.955 was obtained between carbon estimates and total population. A Pearson‟s product moment correlation of r = 0.638 was established between the carbon estimates and SOL. The strong correlations indicated the high predictive power of the disintegrating factors on fossil fuel carbon emissions in Uganda. The Relative Errors of 13.15% in 2000, 13.28% in 2005 and 13.46% in 2010 were obtained in comparing the actual emission statistics with the estimated statistics from the Ghosh et al. (2010) model. A paired sample T- test was done and no statistically significant difference was found between the actual and estimated emissions. The results of the study indicated that it was accurate and therefore Remote sensing technologies have the potential to present efficient approaches for estimating and mapping of fossil fuel carbon emissions within the boundaries of Uganda in a less costly, reliable and repetitive manner while providing consistent quantitative insights into CO2 emissions needed to design practical measures for emission reduction. Further studies should consider the use of transportation data and VIIRS Night Time Light imagery which offer better spatial resolution in order to estimate and map fossil fuel carbon emissions with even better accuracy.