2.1. Study Area
The upper Blue Nile basin originates in Lake Tana and has its outlet at the border to Sudan. The basin covers a large part of the Ethiopian Highlands (175,000 km2
), from an altitude of less than 500 meter above sea level at the Sudanese border to more than 4200 meter above sea level in the centre and the eastern escarpment of the Ethiopian Highlands (see Figure 1
). The climate is dominated by the movement of air masses associated with the Inter-Tropical Convergence Zone (ITCZ). During the dry season from November to March, the highlands are affected by a dry north-eastern continental air mass. From March to May, the ITCZ brings a small rainy season (Belg) to the north-eastern part of the basin. Later in the year, the south-western airstream extends over the entire basin and causes the major rainy season (Kremt) from May to October [34
]. The Kremt accounts for a large proportion of the mean annual rainfall and this proportion generally increases with altitude [15
]. The movement of air masses and the different altitudes are the reason for the different rainfall patterns within the study area. While the northern, western, and southern parts of the study area have one prolonged rainy season from May to September, the eastern part is characterized by a bimodal rainfall regime. To capture these temporal and spatial differences of seasonal climate, the upper Blue Nile basin was split into eight sub-basins of major tributaries with available discharge data, where seven sub-basins where modelled, calibrated, and validated.
• Lake Tana sub-basin
The source of the Blue Nile, the Lake Tana sub-basin, is dominated by a large shallow lake and its surrounding floodplains. These wetlands and several water resource projects, such as hydropower schemes and dams for irrigation and flood control purposes [35
], make modelling difficult, because only little data are available on these natural and human influences. We therefore did not model the Lake Tana sub-basin, but used discharge data from the outflow of Lake Tana from 1982–2010 as inflow to the Upper Abay sub-basin. Since the inauguration of the Tana-Beles hydropower scheme in 2010, the river regime has changed [36
]. However, as no data were available after this change, we were unable to reasonably model more recent runoff.
• Upper Abay sub-basin
The Upper Abay sub-basin is not a tributary, but contains the upstream part of the upper Blue Nile basin between Lake Tana and the Dessie Bridge. Further upstream, no discharge data from the upper Blue Nile River or major tributaries were available. The outlet of Lake Tana was used as inlet discharge for the Upper Abay Sub-basin. The eastern part of the basin is dominated by two rainfall patterns, while the western part has a unimodal rainfall regime. These differences have an impact on plantation activities, and the cropping calendar varies greatly within the sub-basin. For this reason, the cropping calendar of all the watersheds east of the upper Blue Nile and the Beshilo River contains a second crop. In addition to selected CFSR climatic data, we used precipitation data from two observatories of the Water and Land Resource Centre (WLRC) at the eastern borders of the upper Blue Nile basin where data was available for more than 30 years. Average annual precipitation ranges from 500 mm in the northeast to 1750 mm in the northwest.
• Muger sub-basin
More than 75% of the Muger sub-basin is cultivated, mainly with barley and teff. Annual precipitation of only 1200 mm is distributed over two rainy seasons, so we used the cropping calendar from the eastern part of the Blue Nile basin. Recent studies showed that the aquifer system of the Muger sub-basin has a hydraulic connection with the aquifer system of the Upper Awash basin, a basin which does not drain into the Blue Nile [37
• Temcha sub-basin
The Temcha sub-basin is located in the southern Gojam region. More than 70% of the whole sub-basin is cultivated or used for pasture. At 1680 mm, the Temcha sub-basin has the highest average annual precipitation of the whole upper Blue Nile basin. However, the highest measured discharge peaks during the rainy season could still not be simulated with the available precipitation data. Rainfall data originates not only from CFSR, but also from the WLRC observatory at Anjeni [39
• Didesa sub-basin
The Didesa River originates in the Mt. Vennio and Mt. Wache ranges, and is, together with the Anger River, the largest tributary of the upper Blue Nile basin in terms of the volume of water. In the highlands, long-term mean annual precipitation reaches up to 2000 mm, while the lower area receives on average less than 800 mm precipitation per year.
• Dabus sub-basin
The Dabus sub-basin drains the southwestern part of the Blue Nile basin. In its headwater is an area of wetlands of approximately 900 km2
]. The whole sub-basin has a size of 14,700 km2
, over 40% of which is cultivated.
• Beles sub-basin
The Beles sub-basin, located in the western part of the upper Blue Nile basin, abuts the Tana basin and is today linked with the Tana Beles hydropower scheme. With the inauguration of this scheme in 2010, the drainage behaviour of Lake Tana changed and it was no longer possible to model the discharge of the whole upper Blue Nile basin, due to missing data from the outlets of Lake Tana. The size of the sub-basins was reduced to only 3500 km2 and delimited by the Upper Main Beles gauging station, because of inconsistent available discharge data from the main outlet of the Beles River. The small sub-basin has on average 1570 mm precipitation per year, and is dominated by shrubland, grassland, and pasture (70%).
• Lower Abay sub-basin
The lower Abay sub-basin contains the area along the upper Blue Nile basin below the Didesa Bridge, which could not be modelled with larger tributaries. The Upper Abay and all the five modelled tributaries flow into this sub-basin and were included as basin inlets.
2.2. Hydrological Model
For this study, we used SWAT to simulate the discharge of the sub-basins in the upper Blue Nile basin. Other modelled physical processes, such as potential evapotranspiration and base flow [40
], were calculated with the Hargreaves Method [41
] and an automated base flow separation and recession analysis technique [42
], respectively, and used to control the plausibility of the shares of these processes. However, due to a lack of measured data, they could not be calibrated and validated. The model requires input parameters, such as soils, land use, land management, topography, or climate data [43
]. It is designed to calculate runoff and sediments for individual drainage units, called hydrologic response units (HRUs), in generated sub-catchments. It also routes modelled discharge and sediment load towards the outlet of the basin [44
]. A more detailed description of the model can be found in many reviews of its performance and parameterization in Ethiopia and other regions [9