Application of the RUSLE model to estimate sedimentation in the Lwanyo Reservoir in Mbarali District Mbeya – Tanzania

Communities depend on water reservoirs to provide water that supports a range of services including the generation of electricity, water supply for domestic uses, to meet irrigation needs, for industrial uses and recreation (Snyder et al. 2004; Huang et al. 2018). Some of these reservoirs are subject to sedimentation (Wohl & Cenderelli 2000; Huang et al. 2018). Sedimentation occurs when soil is eroded from the land surface or stream banks and is transported by runoff and stream flows into water bodies where the reduction in flow velocity deposits the sediment in water bodies like reservoirs (Tundu et al. 2018). These hydrological processes are complex (Ezugwu 2013). The sediment yield from a catchment, its transportation and deposition are influenced by a range of factors including topography, rainfall, soil types, vegetation and land use.

Several approaches have been applied to estimate sedimentation in reservoirs. In India, the hydrographic surveys are conducted using echo-sounders, satellite-based global positioning systems and computerized methods to estimate sedimentation. This requires extensive fieldwork, costly equipment and skilled manpower (Goel et al. 2002). In contrast, Jain et al. (2002) used remotely sensed data to estimate sedimentation in the Bhakra Reservoir in the Himalayan Region. Heidarnejad et al. (2006) compared estimates of the sediment volumes that accumulated in the Karaj Reservoir using the hydrometry method (a physical approach) and the hydrograph method (an empirical approach). The sediment volume deposited over 30 years was estimated by the hydrometry method and the hydrograph method to be 416,667 and 406,515 m³, respectively. The Lwanyo Reservoir in Tanzania, a critical water source for the Ruanda Majenje Irrigation Scheme, was selected for this study due to its significant role in supporting 2,780 people and irrigating 720 ha of farmland. Unlike other reservoirs, Lwanyo faces acute sedimentation challenges, which have drastically reduced its storage capacity, threatening its ability to meet irrigation demands. Its location in a semi-arid region, coupled with its proximity to highly erodible catchment areas, accelerates sediment deposition, making it a compelling case for studying sedimentation dynamics and their impact on reservoir performance. Additionally, the use of empirical sediment modelling approaches, which require fewer resources while producing comparable results, was preferred for this study to effectively assess the reservoir's sedimentation issues (Jahun et al. 2015a). Lwanyo's socioeconomic importance and susceptibility to sedimentation, combined with limited studies on similar reservoirs in Tanzania, make it an ideal choice for research on sustainable water resource management (Mkhandi et al. 2003; Ngongondo et al. 2011).

Models can be classified based on their complexity and are composed of empirical, conceptual and process models (Ndiritu & Daniell 1999). Process models are designed to use field-measured values of parameters and are intended to closely represent the processes. Most of these models are distributed in order to represent the physical processes and rely on measured values of parameters (Ndiritu & Daniell 1999). Huang et al. (2018) applied physical-based models where the Buckingham π theorem, hydrodynamic similarity and sediment similarity were applied to estimate sediments in the Wushe Reservoir in Taiwan. However, inadequate data, the complexity of the problem and model imperfections limit the application of physical models in real-world problems (Katambara & Ndiritu 2009).

Conceptual models use simple equations to represent the processes, while empirical models do not represent the processes at all. Conceptual and empirical models are calibrated to obtain the model parameters (Ndiritu & Daniell 1999; Katambara & Ndiritu 2009). Generally, the calibration process involves dividing the observed datasets into two sets: one set for calibration and one set for verification (Ndiritu & Daniell 1999; Jothiprakash & Garg 2009; Katambara & Ndiritu 2010). An acceptable objective function is used to minimize the error. The evaluation of model performance attempts to measure the extent to which the model mimics the process (Moriasi et al. 2007; Swilla et al. 2024). Empirical models have been widely used to estimate sediment yield including the Universal Soil Loss Equation (USLE) (Wischmeier & Smith 1978). Improvements made based on 40 years of research have resulted in an improved version of the model, the Revised USLE (RUSLE), which is capable of incorporating distributed data from GIS, is less costly and has an improved accuracy (Jahun et al. 2015b). The RUSLE was adopted for this study.

Several studies have developed site – or region-specific functions to estimate these factors, such as Fu et al. (2006) who developed an equivalent R factor (Re) for the Inland Pacific Northwest. Renard et al. (1997) proposed the erodibility factor, and Moore & Burch (1986) suggested the length of the slope (L) and the steepness of the slope (S). Ghosal & Das Bhattacharya (2020) reviewed the RUSLE model and reported on the variation of rainfall erosivity factor over different climatic zones, the variation of soil erodibility factor over different soil properties, the variation of slope length and steepness factor, the variation of cover management factor and the variation of conservation practice factor. The identification of these variations improved the model's estimation accuracy. The use of the RUSLE model to estimate the sedimentation in the Lwanyo reservoir is described in Section 5.

The Lwanyo Dam is located in the Mbeya Region in southwestern Tanzania and serves as a vital water resource for domestic use, irrigation and agricultural activities (Figure 1). The storage capacity of the Lwanyo reservoir was 210,153 m³. The reservoir's catchment area covers approximately 39.6 km² and is characterized by diverse topographical and climatic conditions. The region receives annual rainfall ranging from 986 to 2,200 mm, with significant seasonal variations that influence runoff and sediment transport dynamics. The Mkoji sub-catchment, which is part of the Lwanyo catchment, experiences an average daily river flow of 1 m³/s during the dry season. Spot discharge measurements fluctuate between 0.06 and 0.3 m³/s, reflecting the semi-arid nature of the region (Mbeya Zonal Irrigation Report 2008).

The altitude within the catchment ranges from 1,350 to 2,734 meters above sea level, contributing to varied ecological and erosion patterns. The highlands are cooler, with average annual temperatures of approximately 16 °C, while the lowlands are warmer, with temperatures averaging 25 °C. These climatic gradients, coupled with the tropical climate in the Igurusi area, create conditions where erosion and sediment transport are influenced by both rainfall intensity and temperature-driven vegetation cover changes. The valley leading to the reservoir spans about 7,284 m, with steep gradients upstream and flatter terrain closer to the reservoir. This topographical variation accelerates flow velocities in the southern, steeper parts of the catchment, increasing the potential for soil erosion and sediment deposition downstream.

The reservoir is situated in a region prone to erosion due to anthropogenic activities such as agricultural land expansion and deforestation, which reduce vegetation cover and exacerbate soil degradation. These factors, combined with the catchment's socioeconomic reliance on the reservoir for irrigating approximately 720 ha of farmland, highlight the critical need for sustainable management practices to mitigate sedimentation and preserve the reservoir's functionality.

Table 1 provides a comprehensive overview of the average monthly temperature (°C) and rainfall (mm) for the Lwanyo Reservoir catchment. The data highlight the climatic variations across months, illustrating both temperature and precipitation patterns critical for understanding the region's hydrology and sedimentation dynamics. The region experiences the highest rainfall in March (239 mm) and December (221 mm), coinciding with a relatively stable average temperature of around 21.8–22.7 °C during these months. These wet months contribute significantly to runoff and erosion, influencing sediment transport into the reservoir. Conversely, the driest months are July and August, with minimal rainfall of 9 and 4 mm, respectively, and cooler temperatures averaging 18.6–19.7 °C. These dry conditions, where there is no rainfall, do not contribute to erosion and sediment transport.

The transitional months, such as May and September, show moderate changes, with rainfall decreasing to 55 and 6 mm, respectively, while temperatures begin to rise, marking shifts between wet and dry seasons. The highest temperatures occur in November (23.7 °C) and October (23.3 °C), indicating a warming trend preceding the onset of the rainy season.

This seasonal variation in temperature and rainfall plays a crucial role in influencing erosion potential, soil stability and the hydrological balance of the reservoir catchment, directly impacting sedimentation rates and water management strategies. Understanding these monthly patterns is essential for planning conservation practices and mitigating sedimentation impacts on the reservoir.

In order to estimate the sedimentation in the Lwanyo reservoir, it was found necessary to base parameter values on the topographical characteristic of the catchment that drains to the reservoir.

It is fair to note that the erosivity factor (R) quantifies the potential of rainfall to generate soil erosion based on intensity and duration (Wischmeier & Smith 1978). High erosivity increases the energy of raindrops, leading to significant soil particle detachment and erosion (Ghosal & Das Bhattacharya 2020). It affects sediment transport and contributes to sedimentation in reservoirs, reducing their storage capacity and operational lifespan, as observed in the Lwanyo Reservoir (Moshi et al. 2024). Regions with high erosivity are particularly vulnerable to land degradation, especially when combined with deforestation or poor land management practices (Tundu et al. 2018). Increased sediment deposition in reservoirs shortens their lifespan and exacerbates flood risks due to heightened runoff rates (Jahun et al. 2015b).

Tundu et al. (2018) report that the conservation factor, P, reflects the control of soil loss by conservation methods. The value depends on whether the area has the maximum conservation or no conservation at all with the value ranging from 0.01 to 1, respectively. Geographic information system (GIS) mapping can be used to estimate the values of the conservation factor.

Using the annual monthly average rainfall of the Lwanyo catchment suggests that the erosivity factor can provide average rainfall energy for various months within a year. As shown in Equation (2), the rainfall erosivity ranges from 0 (MJ mm ha⁻¹ h⁻¹ yr) for the dry month to 421.39 (MJ mm ha⁻¹ h⁻¹ yr). The maximum value is 421.39 (MJ mm ha⁻¹ h⁻¹ yr) in March and the average value is 110.89 (MJ mm ha⁻¹ h⁻¹ yr). In Equation (3), the rainfall erosivity factor ranges from 1.68 (MJ mm ha⁻¹ h⁻¹ yr) to 160.41 (MJ mm ha⁻¹ h⁻¹ yr). The maximum value is 160.41(MJ mm ha⁻¹ h⁻¹ yr) in March. The total annual erosivity factor is 1,330.66 (MJ mm ha⁻¹ h⁻¹ yr) and 960.095 (MJ mm ha⁻¹ h⁻¹ yr) for Equations (2) and (3), respectively. As a conservative estimate, Equation (2) was used.

The vegetation cover and management cover (P) is given in Equation (7). For an average value of c of 39.15%, the value of C is 0.1045.

The catchment upstream of the Lwanyo Dam is a conservation area, and the value of the P factor is adopted from Wischmeier & Smith (1978), as reproduced in Table 3. For a slope of 19%, the value of the P factor is 0.14.

Applying the values indicated in Section 6.1.1–6.1.5 into Equation (1) where factors R, K, LS, C and P are 421.39, 0.071, 10.69, 0.1045 and 0.14, respectively, the value is 4.68 tonnes/ha/year. Moshi et al. (2024) reported that the volume of sediment deposited in the reservoir over an 8-year period was 58,349 m³. Based on an assumed sediment porosity of 0.4 and a sediment density of 1,750 kg/m^3, this equates to a mass of sediment of 61,266 tonnes.

Averaged over the catchment area of 3,960 ha, this gives an average annual sediment export from the catchment of 1.934 tonnes/ha/year. The difference may be attributed to the silt that flows out during the rainy season since the flows are observed to be turbid.

The study, which applies the RUSLE model to estimate sedimentation in the Lwanyo Reservoir, highlights several limitations. Key constraints include reliance on average parameter values (e.g., rainfall erosivity and soil erodibility) that may not capture spatial and temporal variability within the catchment (Ghosal & Das Bhattacharya 2020). The model's inability to explicitly simulate complex sediment transport and deposition processes contributes to discrepancies between modelled (3.977 tonnes/ha/year) and observed (1.934 tonnes/ha/year) sedimentation rates (Moshi et al. 2024). Data gaps, such as incomplete hydrological records, and simplified assumptions about topography and slope uniformity further reduce accuracy (Renard et al. 1997). Additionally, external factors like deforestation, land-use changes and climate variability are not dynamically integrated, limiting the model's predictive utility over time (Tundu et al. 2018). These factors suggest a need for complementary methods to enhance sedimentation estimates.

This study aimed to assess whether the RUSLE model provided a rate of sedimentation that matches the rate of sedimentation observed in the Lwanyo Reservoir in the Mbarali District of Tanzania. The Lwanyo reservoir catchment was calculated to experience erosivity factors (R) with values up to 421.39 (MJ mm ha⁻¹ h⁻¹ yr) in March, a soil erodibility (K) of 0.071, a topographic factor (LS) of 9.086, a vegetation cover and management factor (C) of 0.1045, and a conservation practice factor P of 0.14. The RUSLE Model estimated the annual average soil loss of 3.977 tonnes/ha/year. In comparison, the recorded sedimentation in the reservoir over 8 years, averaged over the catchment area of 3,960 ha, equates to an average annual sediment export from the catchment of 1.934 tonnes/ha/year.

Given that the observed sedimentation has reduced the reservoir capacity by 27.7%, it is estimated that the reservoir would be filled with sediment within a further 20 years in the absence of intervention measures. It is concluded that reducing catchment anthropogenic activities and practising catchment afforestation will reduce soil loss and prolong the life of the reservoir.

The authors acknowledge the support and permission extended to them by the Mbeya University of Science and Technology.

This work was supported by the Mbeya University of Science and Technology as part of postgraduate studies.

B.M. conceptualized the study and wrote, reviewed, and edited the article. Z.K. wrote, reviewed, and edited the article.

All relevant data are included in the paper or its Supplementary Information.

The authors declare that there is no conflict of interest.