Evaluation of sediment generation and transport: a case study of Thwake Dam in Kenya

The problem of sediment deposition and accumulation in reservoirs without a management strategy poses challenges to the long-term benefits accrued from water conservation. In managing scarce water resources, constructing multi-purpose dams ensures regulated and reliable water allocation and efficient use through harvesting and conservation. From a global context, the sediment problem limits the guaranteed sustainable use of reservoirs worldwide due to storage loss. Studies indicate that loss of storage capacity is higher than increased capacity by the construction of new dams.

The Sennar Dam was constructed across the Blue Nile in 1925; a study showed that the sedimentation rate in the reservoir in the period 1925–1981 was 0.5% per year, meaning a 28% reduction in reservoir capacity occurred in the 56 years of operation. Between 1981 and 1986, sedimentation rates increased to 5.8% due to changes in operation rules to satisfy hydropower requirements and irrigation, resulting in a 29% loss in storage in just five years (Gismalla et al. 2015). The loss in storage impacts the dam flood-attenuation capabilities and results in reduced benefits in the long run and eventual decommissioning.

Adjacent catchment data of the Tana Basin in Kenya all estimate sedimentation rates between 350 and 1,000 t/km²/year. High-yielding sediment basins with estimated values over 19,520 t/km²/year were recorded in the Perkerra River basin in Baringo, Kenya (Garde & Ranga Raju 2000). A detailed report for Thwake Dam (SMEC 2018) gave an estimated sedimentation of 569,964 m³/km²/year, which translates to 55–80 t/km²/year. Maruba Dam in the Thwake River basin, constructed in 1950 (Phase 1), had an estimated sediment yield of 650 m³/km²/year from the bathymetric survey conducted in 2008 (NWCPC_NWHSA 2008) and subsequently another study reported an estimated yield of 517 m³/km²/year (Luvai et al. 2022).

Analysis of sediment transport formulas using common equations like Ackers–White, Bagnold, Yang, Colby, Shen and Hung, and Engelund–Hansen were assessed in estimating sediment yield using Euphrates River data from Al Anbar province, Iraq (Sulaiman et al. 2021). The Engelund–Hansen formula showed superior accuracy in predicting sediment loads. The research aimed at enhancing understanding of Euphrates River sediment dynamics, supporting informed decisions on riverbank management and infrastructure development. Annual sediment loads in the Euphrates River within Al Anbar province have ranged from 1.9 × 10⁶ to 2.1 × 10⁷t/year (Al-Ansari et al. 2015). Therefore, the sediment transport equation will be applicable to specific reach conditions and sediment properties. Similarly, cesium-137 was used as a tracer in soil erosion studies due to its widespread distribution in soil and was used as a case study to assess deforestation and its impact on soil erosion (Gharibreza et al. 2020). A radioactive isotope of cesium is a chemical element with the atomic number 55 (Snow & Snyder 2016). It is formed as a byproduct of nuclear fission processes, commonly found in nuclear waste and fallout from nuclear-weapons testing. ¹³⁷Cs emits gamma radiation, making it useful in various applications such as radiological dating and environmental monitoring. By measuring its concentration at different depths within soil profiles, erosion rates can be estimated over time. This involves drying soil samples, grinding and homogenizing, before sieving to separate finer portions, and ¹³⁷Cs activity is then measured by extracting it from soil-core samples and measuring it using gamma spectrometry. Such analysis provides data on soil redistribution dynamics and erosion patterns applicable in areas already exposed to the isotope.

Thwake River exhibits a consistent trend where bed-load transport increases with rising flow and contributions made by ephemeral streams in the transport of bed-load material have indicated the smallest stream power (ω*) value observed in the ephemeral stream is approximately 2.5 times larger than the largest value recorded in the perennial stream (Almedeij & Diplas 2005). In addition, unique bed-load stratification patterns are formed compared with the perennial Athi River, suggesting a continuum in bed-load transport rates with increasing stream power (Almedeij & Diplas 2005). Ephemeral streams demonstrate higher bed-load transport efficiency under similar flow conditions. By applying theories developed for perennial streams, bed-load transport rates in ephemeral gravel streams are simulated, offering practical insights for reservoir management and channel stability assessment.

To estimate sediment transport effectively, a comprehensive approach combining field measurements, empirical relationships, and numerical modeling is employed. During the study, field survey data were gathered, including climate data, hydraulics, sediment samples, and channel characteristics. Empirical equations were used to estimate sediment transport rates. Numerical models to simulate flow, sediment transport, and morphological changes were observed. Calibration and validation ensured accuracy using existing inline structures, while scenario analysis predicted the impacts of deposition over time and the storage life of Thwake Dam.

This study utilizes spatial geographical data and quantitative climate data to develop the numerical model using one-dimensional movable-boundary open-channel flow and sediment simulations using a combination of GIS, HEC-HMS, and HEC-6 RAS Mapper. This integrated method offers valuable insights into sediment dynamics, erosion processes, and sedimentation patterns, guiding river management strategies efficiently. The model output evaluates the impact of sediment generation and transport within the river basin and within the reservoir. The study outcome contributes to sediment management strategy for reservoir conservation and enabling the best decisions on optimum economic operation for maximum benefit.

The initial part of the study involved data collection and desk study to establish hydromet monitoring information, sediment sampling, and analysis. The Landsat data was used for land classification and processed using GIS, online sources from Group on Earth Observations (GEO) and Google Earth Engine (GEE). Hydraulic modeling for sediment transport was developed based on geometric data, flow data, sediment data, and sediment analysis. The model estimates sediment transport potential across selected reaches along the Thwake River basin and upstream of the reservoir from raw data collected, stored, and interpreted.

Climate impact was considered based on climate data from IPCC AR5 SimCLIM and CODEX models (ISC 2020), and UK Met Climate Change models for East Africa 2022. The rainfall datasets comprised hydromet stations identified within the sub-basin with reliable data, which were stations 9137003, Malili Ranch; 9137003, Machakos Agro Met Station; and 9137003, Ngelani Youth. Records from the years 1961, 1962, 1997, and 1998 were used due to extreme recorded storm and available data. Penman–Monteith (PM) reference evapotranspiration (ET₀) values were calculated using the ET₀ calculator developed by the FAO. The calculator uses the minimum and maximum temperatures, mean relative humidity, mean wind speed, hours of bright sunshine, and solar radiation sourced from the CLIMWAT for CROPWAT 2.0 database to calculate the average monthly ET₀ in millimetres per day (FAO 2022b).

The river morphology is examined, specifically the banks, slopes, and sinusoids. Studies on the land-cover data during wet and dry seasons were evaluated using land-cover data obtained from Landsat 4–5 Thematic Mapper (TM).

An event-based model was developed for an extended period to simulate different sediment output values using data on concentrations and flow regimes, among other conditions and controls. The research evaluated sensitivity checks on climate-change parameters and observed impacts on the reservoir. Longitudinal deposition of sediment into the reservoir is a phenomenon referred to as delta deposit.

The primary impact of these delta accumulations is the elevation increase in backwater levels within the channel upstream of a reservoir. Consequently, the delta formation can introduce flood risks that may not have been foreseen during the design phase. Sediments deposited in the delta undergo continuous reworking into the downstream storage region, particularly during periods of low-reservoir stage and extreme flood discharges. The delta forms based on particle sizes and deposits in the extreme upstream portion of the profile. Although the delta study is used in defining additional inundated areas caused by backwater, it is critical in managing an advancing delta in the reservoir safe storage areas as explained in subsequent sections.

Sediment discharge and prediction of deposition into the Thwake reservoir were computed using HEC-6, a one-dimensional movable-boundary open-channel flow model that computes sediment scour and deposition by simulating the interaction between the hydraulics of the flow and the rate of sediment transport. The model incorporates the assumption that equilibrium conditions are achieved between the flow and the bed-material transport within each time-step, an assumption also made in most other sediment transport models (Morris & Fan 1998). This assumption may be violated during rapidly rising and falling hydrographs, which can limit the model's ability to simulate single events. For this reason, the HEC-6 program documentation specifically states that the model is designed for the analysis of long-term river and reservoir behavior. However, in practice, it has also been used to simulate single events (HEC 2022).

The various sediment concentrations were generated based on sediment yield values and sampled data, and the graphical data obtained is shown in Figure 3. The sediment concentration values were based on flow profiles simulated using results from the HEC-HMS model for extended periods with data from the El Niño events of 1961–62 and 1997–98.

Sediment erosion and transport were iteratively calibrated against the HMS runoff values with different sediment concentration values. Since there is no regular gauging station in sub-basin 3E, correlation between RGS 3F02 (1 km downstream of confluence) and RGS 3DA02 (upstream of confluence) was plotted and a mass balance developed with discharge from 3E.

The bathymetric survey information was relied based on data from Maruba Dam located within sub-catchment 3E and Masinga Dam in the adjacent catchment. Applying the climate factor to the model was compared with estimated sediment yield results from dams in semi-arid areas in USA sediment (Brandt et al. 2017). The soil moisture accounting (SMA) loss method included in HEC-HMS was employed to model infiltration losses combined with canopy and surface methods.

With data from the existing Maruba Dam and applying climate-change impact with a 15% increase in precipitation (ISC 2020), the estimated yield is computed at 530 m³/km²/year (Luvai et al. 2022), equivalent to 1,405 t/km²/a and 0.724% sediment concentration (Scenario 1). The mass/volume budget gives mass cumulative sediment inflow into the reservoir estimated at 15.8 metric tons of sediment inflow. This represents 2.3% reservoir loss and a useful life under 43 years.

Scenario 2 estimated yields computed at 475 m³/km²/year, equivalent to 1,260 t/km²/a and a 0.65% sediment concentration. The mass/volume budget gives mass cumulative sediment inflow into the reservoir estimated at 14.24 metric tons of sediment inflow. This represents 2% reservoir loss and useful life under 50 years.

Scenario 3 estimated yields computed from RUSLE at 878 m³/km²/year, equivalent to 2,327 t/km²/a and a 1.2% sediment concentration. The mass/volume budget gives mass cumulative sediment inflow into the reservoir estimated at 26.3 metric tons of sediment inflow. This represents 3.8% reservoir loss and a useful life under 26 years (Table 1).

Sediment discharge and prediction of deposition into the Thwake reservoir were done using HEC-6, a one-dimensional movable-boundary open-channel flow model, by computing sediment scour and deposition and simulating the interaction between the hydraulics of the flow and the rate of sediment transport. This aspect is fundamental in managing sediment deposition and reservoir operation (Morris & Fan 1998). The model operates on the premise that a state of equilibrium is achieved between the characteristics of the flow and the movement of bed-material transport during each discrete time-step (HEC 2022). Notably, this assumption mirrors the foundation upheld by numerous other sediment transport models that also hinge upon the attainment of equilibrium conditions.

Therefore, mitigation measures are needed to extend the life of the dam beyond 15–26 years as predicted by the scenarios simulated, especially when considering future climate-change impacts. Several measures are proposed, both social and technical, which will be spearheaded by the Water Resources Users Associations (WRUAs) and responsible institutions.

The findings indicated that sufficient bed material from sub-basin 3E would be deposited into the reservoir, resulting in delta formation approximately 5 km downstream of the tail waters at minimum dam operating level. The mass cumulative sediment inflow from 3E into the reservoir was estimated as between 14 and 26.3 metric tons per annum, representing reservoir loss and useful life under 50 years.

The construction of a submerged check dam is proposed at the delta point to limit progress towards the intake facilities at the embankment of the reservoir. Conservation measures within the sub-basin are recommended such as sand dams. Sand dams, once silted up, can increase the amount of arable land in the region 15-fold (Maddrell 2018). They are each typically about 2 m high, 40–50 m long, and provided with a spillway. The check dams should trap sand but mostly silt based on the hydrological modeling of the sediment routing. The construction of sand dams is suggested to reverse and prevent desertification/land degradation and to mitigate the effects of drought in affected areas to support poverty reduction and environmental sustainability. These interventions can guarantee over 50% reduction in sediment yield translating to the full design life of the reservoir.

The successful adoption of conservation practices is contingent upon a comprehensive consideration of not only their technical efficacy but also the prevailing socioeconomic dynamics. Particularly in the context of the project area, non-governmental organizations (NGOs) have demonstrated adeptness in devising strategies that effectively engage rural populations. In contrast, governmental agencies have at times yielded sub-optimal outcomes. The following features are hallmarks of a successful WRUA participation in the conservation.

The conservation measures proposed are dependent on the slope, current land use, rainfall distribution according to agro-climatic zones, and also whether the conservation works are to be carried out on-farm or off-farm. Hotspots within the catchment to yield an associated reduction in sediment yield were evaluated and it was concluded that 1,400 km² of sub-basin 3E and 3D need intervention to yield a 15%–23% reduction in sediment yield translating to 6–12 years' additional reservoir life (Kiringu et al. 2022). Continuous sediment measurements flowing into and out of the reservoir are critical. Remote measurements of sediment loads or concentrations by traditional methods can be expensive and are generally infrequent. Turbidity hydro-acoustic or optical instruments can be used continuously to measure sediment loads entering or leaving the reservoir. These methods are less expensive than direct measurement methods and provide high-resolution time-series data. The advantage of acoustic measurements is reliable data for proper decision-making such as upscaling upstream interventions. Such phenomena as density current/turbidity current may be recorded by the sensors.

The sediment release efficiency of a reservoir is the mass ratio of the released sediment to the total sediment inflow over a specified period. It is the complement of trap efficiency. Empirical relationships have been found to provide reasonable estimates of long-term releasing or trapping efficiency. The effects of reservoir operation are included only to the extent that they are reflected in the selection of the pool volume used in the computations. Judgement is required to adjust these methods to specific conditions (Morris & Fan 1998).

Thwake River exhibits a consistent trend where bed-load transport increases with rising flow; contributions made by ephemeral streams in the transport of bed-load material have indicated the smallest ω* value observed in the ephemeral stream is approximately 2.5 times larger than the largest value recorded in the perennial stream (Almedeij & Diplas 2005). In the estimation of bed-load transport by desert flash-floods (Reid et al. 1996), bed-load sediment transport equations were assessed using field data from desert wadi flash-floods. The Meyer-Peter and Müller equation performs well with median calculated to observed (C/O) ratio of 1.18). In contrast, a study on sediment transport dynamics of the Euphrates River at a thermal power station in Al Anbar province, Iraq (Sulaiman et al. 2021), indicated that the Engelund–Hansen formula provided the most accurate predictions for this specific river reach.

The case of Sennar Dam, Sudan, where 29% loss in storage was recorded in just five years (Gismalla et al. 2015), points to the importance of reservoir operation rules in sediment management. In consideration of erosion on degraded land, a study of the Hyrcanian forests in Iran (Gharibreza et al. 2020) showed 100% efficiency in the restoration of carbon stock over the 30-year span of the reforestation plan that points to the importance of environmental intervention.

This research combines both basin characteristics and application of appropriate transport function in a numerical model to estimate sediment generation, yield, and deposition in the Thwake reservoir. The study contributes to management strategies and understanding of sediment transport dynamics. The estimation of sediment generation and transport using numerical techniques provides flexibility and reliability. Thwake reservoir may not have the 100-year design life as anticipated if there are no interventions on a sediment management plan. All strategies pertaining to sediment reduction should be considered in this case including environmental, social, and structural/physical interventions. Different sediment predictors point to high sediment-load trapped by the reservoir in less than 50 years, estimated in this study at between five and ten years in the worst-case scenario depending on climate predictions.

The financial implications may be unbearable to stakeholders and operators. The immediate concern is to reduce sediment accumulation by carrying out immediate and short-term actions for reservoir-life elongation. This will involve reducing the sediment inflow into the reservoir by putting in place measures to attenuate storm hydrograph and limit conditions for sediment (silt–sand) generation and transport through sediment monitoring plans.

The authors wish to thank the individuals and institutions that contributed to the completion of this research paper.

Data cannot be made publicly available; readers should contact the corresponding author for details.

The authors declare there is no conflict.