Integrated impact quantification of proposed hydropower plants in Africa

Underlying hydropower plant data

Data on existing and proposed HPPs in Africa were obtained from the Renewable Power Plant Database for Africa (RePP Africa)6. This database includes HPPs with a capacity larger than one MW and differentiates between the reservoir, run-of-river, and pumped storage operating types. For complementing the HPP characteristics, projected reservoir areas for the proposed HPPs (types reservoir and unknown) were estimated with the Python Toolbox DamTools for ArcGIS Pro38. For the reservoir estimation, the flooding of the upstream digital elevation model (DEM) at a dam position is simulated up to the reservoir water level that is limited by the dam height. The simulation uses data on river segments from the RiverATLAS39, major river basins from the HydroBASINS database40, simulated annual mean natural discharge data with 15″ resolution from the Water GAP 2.2 model41, and the SRTMGL3 3″ DEM42. The reservoir delineation of the HPPs as included in the Renewable Power Plant Database (RePP) Africa resulted in two datasets for further analysis: Dataset 1 includes 507 HPPs of type reservoir and type unknown for which reservoir delineation was successful. Seventeen plants out of the initial 524 HPPs were excluded due to failed reservoir delineation. Four of these 17 plants are projects with capacity updates or other HPPs that operate at existing reservoirs and will not cause an inundation of new areas. We tested sensitivity by running all calculations excluding HPPs of type unknown, further referred to as dataset 2 with a total of 236 data entries. Seven HPPs were excluded out of the initial 243 entries due to failed reservoir delineation (Tab. S1).

Since impact quantifications are to be conducted on a river system scale, some of the selected impact indicators as outlined below require details on dam location and reservoir area also for already existing HPPs and existing other dam infrastructure (e.g., for irrigation or drinking water purposes). RePP Africa includes information on reservoir areas for HPPs only; therefore, we additionally used data on existing dams and reservoirs with main purposes other than electricity generation from the Global Reservoir and Dam Database (GRanD))43.

Indicator selection

Indicators for potential HPP impacts were selected based on an approach introduced by Peters et al.6: First, the set of selected indicators represented measurable ecological and socioeconomic variables to underpin the idea of quantifying potential consequences of dam locations in the frame of sustainability. We do not add the dimension of changing environmental conditions over time to treat all indicators equally, i.e. they reflect the current knowledge and available data, and to avoid a bias in including additional uncertainty resulting from assumptions about e.g. hydrological dynamics with high temporal resolution. Second, comprehensive and high-resolution spatial data were available via open-access platforms to ensure transparency and reproducibility.

Therefore, the following eight indicators were selected exemplarily for this study:

1. River Regulation—change in average annual river flow due to dam building; given by an index in % (ecological indicator) 2. River Fragmentation—degree of fragmentation of the river network into smaller river stretches due to dam building; given by an index in % (ecological indicator) 3. Sediment Entrapment—retention of river sediments in the reservoir due to the dam itself and the reduced water flow velocity in the reservoir (compared to a river); the indicator value is given in the resulting lifespan of the reservoir in years, i.e., the number of years until the reservoir is filled up with sediments if no sediment removal is implemented (ecological and economic indicator) 4. Megafauna—overlap of proposed HPP dam sites with IUCN data on areas of presence of freshwater megafauna species as an indicator for the potential threat of dam construction and reservoir inundation on biodiversity; given in species number (ecological indicator) 5. Protected Area—overlap of the projected HPP reservoir with protected areas as an indicator for potential HPP impacts on areas of “high ecological value”; given in square kilometres (ecological indicator) 6. Land Use Change—overlap of the projected HPP reservoir with cropland to indicate a potential conflict with food production; given in square kilometres (social and economic indicator) 7. Resettlement—potential resettlement of people living in the area of the HPP reservoir to be flooded; given in number of people (social indicator) 8. Potential Evaporation—change in potential evaporation due to turning a flowing river into a (almost) standing water with a large water-atmosphere interface that favours evaporation; given in cubic kilometres per year (ecological indicator)

While the indicators Megafauna, River Regulation, River Fragmentation, and Sediment Entrapment measure effects that can be traced back directly to the construction of an HPP (including a dam wall as a barrier), the indicators Protected Area, Land Use Change, Resettlement, and Potential Evaporation depict indirect effects caused by the flooding of an area (projected reservoir area).

Depending on the indicator, alterations accumulate and/or are delayed in time, causing synergistic effects on larger spatial and temporal scales. All calculations to quantify the selected impact indicators for each proposed HPP in both datasets were carried out and their results visualised with ArcGIS Pro 2.9.444.

Quantification of potential impacts of proposed hydropower plants according to selected indicators

For the indicators River Regulation and Fragmentation, the River Regulation Index (RRI) and the River Fragmentation Index (RFI) were calculated, respectively. RRI and RFI include the current degree of regulation and fragmentation caused by existing HPPs and other dam infrastructure. Both indices were calculated by applying a methodology developed by Grill et al.45. The basis for these calculations is data from the HydroSHEDS database46 and the HydroBASINS dataset (Level 3, Level 4)40. Discharge data is obtained from the WaterGAP model41. To calculate the number of years until the reservoir would be entirely filled with catchment-eroded sediments (Sediment Entrapment), a revised version of the Universal Soil Loss Equation (USLE) considering the entire basin with existing reservoirs of all purposes and projected HPP reservoirs was used. The USLE, an empirical model designed to predict the annual average soil erosion from runoff using information on soil properties, rainfall, topography, and land use, is used in combination with a geographic information system (GIS)-based Index of Connectivity (IC) to estimate the fraction of catchment-eroded material that is likely to reach the fluvial network and ultimately be trapped behind the dam wall47,48. The IC integrates an upslope component and a downslope component. The upslope component represents the erosion potential of the contributing area as a function of catchment area, average slope gradient and land-use impedance (USLE C-factor). The downslope component represents the flow-path-weighted distance to the nearest sink, here defined as the dam location. This combined approach thus estimates sediment delivery to each proposed reservoir, rather than full catchment-scale sediment transport. Existing dams and reservoirs from the GRanD database were included into the IC calculation as intermediate sinks within the flow network, thereby accounting for the sediment-trapping effect of upstream infrastructure on the connectivity of eroded material to downstream proposed HPPs49. This combined USLE–IC approach targets the physical soil-erosion process and the connectivity of eroded material to the reservoir; it does not simulate full catchment-scale sediment transport or yield and neglects instream sediment dynamics, and the resulting values therefore carry greater uncertainty than the flow-based indices RRI and RFI. Accordingly, the indicator is intended for the comparative ranking of proposed HPPs rather than for the precise quantification of sediment volumes delivered to each reservoir50.

Impact calculations of all other single indicators were based on projected reservoir areas and considered proposed HPPs only. To estimate the number of megafauna freshwater species (≥ 30 kg, Megafauna), including reptiles, amphibians, and mammals that might be affected by dam construction, we intersected data on the presence of freshwater megafauna with the georeferenced dam locations of proposed hydropower plants12,13. For the indicator Protected Area, projected reservoir areas were intersected with protected areas using a polygon layer from the World Database on Protected Areas51. For the indicator Land Use Change, we used land cover maps from 2015 (Land Cover Maps v2.0.7) provided by the European Space Agency’s Climate Change Initiative with a resolution of 300 m × 300 m52. We extracted cropland as an indicator for food production and overlaid it with the projected reservoir areas.

In order to estimate the number of people that must be resettled from potentially inundated reservoir areas (Resettlement), we overlapped projected reservoir areas with a population raster dataset obtained by the WorldPop Project with a resolution of 3 arc seconds53. For the indicator Potential Evaporation, the potential evaporation values [mm/year] were obtained from the FAO GeoNetwork database with a resolution of 10 arc minutes54. Evaporation values of the raster feature were bound to reservoir points, and potential evaporation [m3/year] was calculated by multiplying each reservoir area [m2] with its respective evaporation value [mm/year].

Correlation between independent indicators and projected reservoir areas was tested with R using the rank correlation coefficient Spearman’s rho. This allows us to analyse a potential bias of the ranking of the indicators towards an overlap of the indicators in their interpretation, e.g. “Protected Area” and “Megafauna” as indicators for biodiversity.

Ranking of proposed hydropower plants considering all indicators and indicator compositions

The impacts of proposed HPPs in Africa were quantified for dataset 1 (HPPs of type reservoir and type unknown), and dataset 2 (HPPs of type reservoir), following the methodological approach summarised in Fig. 2. First, indicators of ecological and socioeconomic relevance were selected and analysed separately. The proposed HPPs were ranked for one indicator at a time according to their calculated impact, and quartiles were determined. Second, following the ranking of each single indicator and the resulting quartile values, every HPP was given the value of the respective quarter they belonged to: from one representing low impact; two, moderate impact; three, heavy impact; to four, severe impact. Third, quarter values were summed for each proposed HPP. Every proposed HPP was ranked according to its overall impact (sum of quarters) (1) for the entire continent and (2) within major basins (Congo, Niger, Nile, Volta, Zambezi). Weighting of indicators could be integrated at this point if stakeholder opinion or discussion outcomes of (local) representatives would support emphasizing of specific indicators, but was neglected here to avoid additional bias. Because impacts of large environmental projects occur in various sectors on different spatial scales and across time, it can be crucial to test if a change of a certain input parameter or an input parameter composition changes the outcome55. In order to test the sensitivity of the model to the composition of the indicators, analyses were re-run in a “leave-one-out” fashion (one indicator was excluded at a time, local sensitivity approach55,56), and then again through excluding several indicators by creating two different scenarios (Scenario A: local-scale indicators; Scenario B: catchment-scale indicators). After the exclusion of one or more indicators at a time, rankings for both spatial scales, (1) continental and (2) major basins, were created. Results of the impact quantification considering single indicators, all indicators, and indicator compositions were compared among each other (Fig. 2) and visualised with ArcGIS Pro44.

Fig. 2 Full size image Methodological approach: Impact quantification of proposed hydropower plants and ranking for every single indicator (1) and for different indicator compositions (2). All calculations were run twice: Dataset 1 includes all proposed hydropower plants (HPPs) of type reservoir and type unknown, and dataset 2 includes all proposed hydropower plants of type reservoir. Data on hydropower plants was obtained from the Renewable Power Plant Database (RePP) Africa6 and the Global Reservoir and Dam (GRanD) Database49. Additionally, we used data on existing dam and reservoir infrastructure for hydropower and other purposes from GRanD49.

We created two different indicator-composition scenarios to better understand if and how the selection of indicators might change the ranking of the proposed HPPs. Scenario A includes the indicators Protected Area, Resettlement, Land Use Change, Megafauna, and Potential Evaporation, which assess the direct effect of reservoir inundation and dam construction on a local scale. Scenario B includes the indicators River Regulation and Fragmentation as well as Sediment Entrapment, all of which are consequences of dam construction. All three indicators assess impacts that cause cumulative system alterations at a large scale, and their calculation considers basin-wide effects of existing dams as well as the complementary building of new ones. The created scenarios are examples that intend to show that rankings at basin or continental level generally depend on the selection of indicators.