Variability of Rainfall and River Flow in the Nile Basin

Rainfall variability engenders changes in the frequency and severity of water-related disasters. In the Nile basin, rainfall-based disasters e.g. landslides, flooding and dry conditions are common. This research aimed at providing a platform for insights on variability of historical rainfall and flow in the River Nile basin. Spatio-temporal variability of both rainfall volumes and extreme events across the study area was analyzed in a quantile-based way. The possible drivers of the rainfall variability were investigated. The influence of spatio-temporal scales on rainfall variability analysis was investigated using the empirical orthogonal functions. The main driver of flow variability was investigated in both data and simulation-based ways through assessment of rainfall-flow co-variation and rainfall-runoff modeling respectively. Uncertainty in sampling of hydro-climatic quantiles was assessed using extreme value analysis.

Both rainfall totals and extreme events exhibited oscillatory behavior over multi-decadal time scales across the  study area including the Southern (SOU), Central (CEN) and Northern (NOR) regions. Rainfall variability possible drivers were found to be the variation in: Sea Surface Temperatures (SST) of the Atlantic and Pacific Oceans (for CEN), sea level pressure of the North Atlantic Ocean and the SST of the Indian Ocean (for SOU), and the SST of Indian and Atlantic Oceans as well as from the Pacific Ocean (for NOR). The amount of explained variability was found to reduce with the increase in temporal resolution or spatial domain size. Rainfall variability was found to be more suitably explained at a regional than location-specific spatial scale. The largest amount of river flow variation over the data period 1950-2000 was found to be explained by only the rainfall variability. The null hypothesis H0 (no correlation between rainfall and flow) was rejected at the 5% significance level for several catchments. The sampling uncertainty in terms of the difference between the limits of the 95% confidence intervals expressed as a percentage of the empirical 10-year daily quantile was, on average, 35% (for rainfall intensity), 22% (for high flow), and 51% (for low flow).