International development projects are a massive business, with billions invested annually in the Global South. However, such projects have an unacceptably long record of high failure rates. The problem perpetuates, in part, due to the success factors by which international development projects are judged. Often, projects are assessed on the basis of donor-identified priorities that are not aligned with local impacts. One such international development project involves the construction of small-scale water harvesting structures known as sand dams. Non-governmental organizations (NGOs) continue to raise sufficient funds to build thousands of sand dams across sub-Saharan Africa, and yet 50% of sand dams are estimated to be non-functioning.
Sand dams are small, reinforced concrete dams built across an impermeable stream-bed. Over time, sand settles behind the dam, creating an upstream sand reservoir that fills with rainwater and surface runoff. The sand helps filter the water, protects it from evapotranspiration, and can provide water to the local community for domestic and agricultural use during the dry season. Sand dams often fail due to poor construction, inadequate siting, and siltation.
This dissertation explores methodologies for studying the regional and local impacts of sand dams and investigates the feasibility of developing model-based site selection guidelines for sand dams. Three objectives of this study are: (1) to develop a methodology to assess the ability of sand dams in improving the overall water availability in the region; (2) to examine claims made by non-scientific bodies about sand dam impacts by investigating how diverse sand dams influence macroinvertebrate habitat, vegetation, erosion, and local water availability; and (3) to create guidelines for siting new sand dams based on a fully integrated surface and groundwater flow model.
For the first objective, two multiple regression models are developed to compare (1) water storage and (2) vegetation in an area with a high density of sand dams, termed the sand dam counties (SDC), to those in a control area. The models analyze remotely sensed datasets to assess whether evidence exists of significantly increased storage in the SDC relative to the control area. The results show that the remotely sensed water storage data is unable to consistently detect higher levels of water storage in the SDC. This is likely due to the low resolution of the dataset combined with the small magnitude of sand dams' impact on regional water storage. The results of the vegetation model show that the sand dams have a consistent, positive impact on vegetation within the SDC relative to the control area. Because vegetation health and cover is often correlated with groundwater levels, these results likely indicate that the sand dams are also increasing local groundwater levels. Overall, this study shows that remotely sensed dataset can provide a useful basis to assess the impact of international development projects, particularly those that involve the natural environment.
For the second objective, data relating to macroinvertebrates, vegetation, erosion, and water table elevations at three sand dams were collected and analyzed during a year-long field study in Tanzania. These study subjects were specifically selected to test an NGO claim that sand dams revitalize the entire ecosystem. The results of this study show that sand dams are not a suitable habitat for macroinvertebrates due to their homogeneity. The impact of sand dams on vegetation cover can be significant, but may be limited by the slope of the surrounding land. Functioning sand dams likely have little impact on streambank erosion, but non-functioning sand dams may contribute to the erosion of streambanks in unstable reaches. Lastly, the water table is locally raised by recharge from sand dams, however, the spatial and temporal extent of the impact is more limited than conveyed by NGOs and previous studies. This study adds to the limited body of knowledge on the environmental responses to sand dams and demonstrates the importance of examining the local impacts of individual international development projects.
For the third objective, results from four different simulations of a watershed-based model with three cascading sand dams are analyzed to identify overland features that improve vadose zone storage and groundwater recharge and reduce evapotranspiration. Results from this study show that sand dams constructed in a low-lying area that collects surface runoff from adjacent steep slopes, such as in a U-shaped valley, will likely collect and store sufficient water for use by a local community. Watersheds with relatively more area cultivated with low-water-need crops will similarly be beneficial to sand dam performance. In addition, the analysis revealed that the volume of water a sand dam receives during a rainy season is less important for water storage than the duration of dry seasons. Lastly, the simulations showed that sand dams constructed in an area with sandier soils will perform better than those in an area with loamy soils. This study produced a set of guidelines that can be used to identify locations where sand dams are likely to capture and store sufficient water for community use during the dry season.