Influence of climate variability on representative QDF predictions of the upper Blue Nile Basin

Investigation was made on the hydrological extremes of the upper Blue Nile basin in Ethiopia through the use of high and low flow probability distributions and flow-duration-frequency (QDF) relations. The dependence of QDF predictions on the period used for the analysis was investigated considering the possibility that short historical records might cover periods with a cluster of wet/dry years or where there was a high/low climate oscillation. In such cases, the derived relationships might not be representative of the long term natural variability and might lead to significantly different (biased) predictions. The results show that the high and low flow extremes of the basin could be best described by exponential and Fréchet distributions, respectively. The 1960s–1970s and the 1990s–2000s were identified as wet periods while the 1980s had more dry years. The constructed QDF curves using wet and dry periods reveal significant variations on their estimation specific to the sub-basins. For the entire upper Blue Nile basin, the wet/dry periods could be evaluated relative to the long term 1964–2009 data available at the downstream location, El Diem station. High flow QDF estimations using wet periods are in similar order of value with the long term QDF predictions. The exception is when data concentrates on the dry period (1980s); then the QDF statistics are not representative. It was found that bias correction of around 15% in the flow quantiles is required to shift the QDF statistics to the long term values. The required shift could be fully explained by the climate oscillations. On the contrary, low flow QDFs show shifts that do not necessarily correspond to the changes observed in the climate; rather they are influenced by combined effect of changes in climatic variables and changes in land/water management practices. Since high flow extremes appear to have link with large scale atmospheric variables, it would be interesting to further investigate how the temporal changes in high flow statistics can be linked to the atmospheric variables.status: publishe