Application of a hydrological model in a data-poor tropical West African catchment: a case study of the Densu Basin of Ghana

This thesis reports on the suitability of using a hydrological model to predict the impact of future catchment changes on the surface water resources of tropical West African catchments, as a basis for sustainable watershed management strategies. The economically-important but data-poor Densu Basin (2100km2) in Ghana is used as a case study. Prior to the identification of an appropriate model, a range of model selection criteria to achieve the research objectives were developed, which included minimum data requirements. Following a review of available models, the ACRU hydrological model was chosen using these selection criteria. As this is the first use of the ACRU model in the Tropics, a sensitivity analysis was performed to gain a better understanding of the correspondence between the data and the physical processes being modelled and to inform future data collection priorities. The most sensitive parameters identified were rainfall; soil water content at field capacity and monthly crop coefficients. The sensitive parameters were different from those previously identified for semi arid regions, thereby improving the wider understanding of the behaviour of the ACRU model in a new climatic region of Africa for future studies. Although the ACRU model performed well during calibration (e.g. Nash and Sutcliff Efficiency > 0.8), it performed poorly during the validation period (e.g. NSE < 0.5). Aggregation of the daily output to monthly averages improved the performance appraisal statistics to a level where the model is appropriate for longer term water management issues such as irrigation planning and water supply planning. However, the prediction of the hydrological consequences of future catchment changes using the model could not be performed due to the variable model performance. A detailed analysis of input data quality and model performance identified a number of contributory reasons which included the poor distribution of available data from rain gauges and evaporation stations, poor data management and problems with groundwater processes within ACRU. Recommendations to the Densu River Basin Management Board to address the main problem area of data quality include ensuring that all data (new as well as existing) is subject to QC/QA procedures, the development of data archiving / back-up systems, making historical data available to the wider hydrological community; and guidance on deriving model inputs from available data for future modellers. Finally recommendations on which areas need further study during future research using the ACRU model are given.