Estimation of evapotranspiration and its relationship to land use and water resources in Lufumi Catchment, Kinshasa, D.R.Congo.

Abstract

This study was carried out in an ungauged catchment of the lower Congo River basin, with the overall objective of deriving useful data in support of water resources management practices. Specifically, the study aims to generate and validate remotely sensed water resources data at the catchment level; use this data and assess spatial and temporal changes in water resources and land use at the catchment level; and assess the relationship between the catchment evapotranspiration and varying types of land use. The methodology used in this study consisted of delineation of the catchment, extraction of water resources data from remote sensing products such GLDAS, CHIRPS, for rainfall and GLDAS, MOD16, and FEWS NET for evapotranspiration. The study used empirical approaches such as the Turc method to validate remote sensing evapotranspiration products and Soil Conservation Service Curve Number method to compute runoff of the catchment. Rainfall from the ground were used to validate remote sensing rainfall products. The supervised classification method and Maximum Likelihood Algorithm were used for the classification of land use land cover. The supervised classification was validated using the overall accuracy and Kappa coefficient equations. It was observed that the remote sensing rainfall products were strongly correlated to ground measurement at the catchment level (r=0.870) with p-value<0.05. The remote sensing evapotranspiration time series have the same pattern and the GLDAS actual evapotranspiration remote sensing product was strongly correlated to the potential evapotranspiration computed during the study period (r=0.629) with p-value<0.05. The trend of runoff was found closely related to that of rainfall (r=0.97) with p-value<0.05. It was observed that the ETa and temperature trends have the same pattern but not strongly correlated (r=0.39) and not statistically significant (p-value<0.05). Furthermore, the inter annual variability analysis indicates similar behavior for rainfall and runoff for the catchment. It was observed that a year to year relationship between ETa and runoff is not consistent (r=0.36) p-value>0.05. In general, there was an increase in areas covered by cultivation p-value<0.05 and settlement and bare land but statistically insignificant for p-value<0.05. The decrease in the area covered by savanna p-value<0.05, forest p-value<0.05, and water and marshy p-value<0.05. The overall accuracy of classification was found to be 73% and Kappa coefficient 68%. The study shows that the increase in cultivated areas led to increases of actual evapotranspiration. The reduction of forest and savanna areas contribute to the reduction of actual evapotranspiration in those areas. This led to the reduction of moisture supply into the atmosphere and reduce local rainfall at the catchment level. It was seen that settlement and bare land areas increased, but the actual evapotranspiration was still increased as well, this can be put on the back of increased evaporation from the soil. It can be realised that land use practices have limited impact on actual evapotranspiration. This study reveals that the change in actual evapotranspiration for the Lufumi catchment is more related to hydro-meteorological parameters than observed rates of changes in land use and land cover.