Comparative performance of rural water supplies during drought

The study addresses how different rural water supply technologies perform during severe drought in Ethiopia, with the overarching question of which source types provide the most reliable and resilient service under climate stress. Against a backdrop of increasing drought frequency and severity in sub-Saharan Africa due to climate change, Ethiopia experienced an extreme 2015–16 drought, exacerbated by a strong El Niño and failure of the JJA rains. This crisis affected more than 10 million people and highlighted vulnerabilities in water access, especially from shallow and surface sources. While groundwater is generally resilient, few studies have compared the performance of different rural technologies during drought. Using weekly monitoring of over 5000 water points across diverse altitude and hydrogeological zones, the study evaluates functionality, usage, travel times, and user perceptions for hand-pumped boreholes, motorised boreholes, protected springs, protected hand-dug wells, and open sources to inform strategies aligned with SDG targets for safe and reliable water.

Prior work indicates that groundwater tends to be more resilient than surface water during drought in sub-Saharan Africa, though evidence comparing specific rural technologies is limited. A recent field study suggested hand-pumped boreholes are more resilient than springs and hand-dug wells. Broader literature documents increased drought frequency/magnitude in Africa, Ethiopia’s high vulnerability, and the implications of El Niño events for drought risk. Under the MDGs, many countries expanded access to improved point sources, but some technologies (springs, hand-dug wells) may be unreliable in drought, susceptible to contamination, and have higher life-cycle costs than hand-pumped or motorised systems. The SDG goal for safely managed water emphasizes on-premises access, availability, and quality, but provides limited guidance for national customization in drought-prone rural contexts. Studies also highlight the importance of operation and maintenance capacity, the limitations of community management alone, and the benefits of multiple-source portfolios to enhance resilience and meet varied household needs across seasons and uses.

Design: The study analyzed performance of five rural water source types common in Ethiopia: boreholes equipped with hand-pumps, motorised boreholes, protected springs (including gravity-fed supplies), protected hand-dug wells, and open sources (unprotected wells and surface water). Emergency water trucking introduced during the drought was also tracked for access/usage comparisons. Each source type typically taps different depths and hydrogeological zones: shallow groundwater (0–15 m: hand-dug wells, springs, open sources), intermediate (15–30 m: hand-dug wells, hand-pumps), deep (>30 m: hand-pumps, motorised boreholes), and surface water (open sources). Sampling and period: A census-style baseline inventory visited 5196 individual sites across target Woredas. Weekly monitoring then followed a sub-sample from January to May 2016; 3188 sites were visited more than once (mean 3.5 visits over 12 weeks). A household survey was conducted in the final month to quantify volumes collected. Data collection: Trained enumerators interviewed water point attendants, WASH committee members or other knowledgeable informants using Akvo Flow mobile surveys (Samsung Galaxy J1, Android 4.4.4). Data were stored on devices and uploaded when connected. Functionality definition: Three user-reported categories were applied and visually confirmed by enumerators: functional (operating as usual), partially functional (reduced performance but some water), and non-functional (no water at the time of visit). These categories reflect perceived functioning rather than measured yield or detailed reliability metrics. Metrics: Weekly averages were computed for functionality rates (share of points by category), user numbers per point, and modal travel times to collect water. User perceptions of sufficient quantity (yes/no) and, from the household survey, modal litres per capita per day (lpcd) were analyzed. Context stratification: Ethiopia was partitioned into three altitude zones that influence rainfall, water availability, and demand: Kolla (<1500 m, lowland, high variability and recurrent drought), Weyna Dega (1500–2400 m, mid-altitude, high population and demand), and Dega (>2400 m, highland, lower temperatures and lower demand). Analyses were stratified by these zones. Climate context: Rainfall climatology and anomalies were characterized using TAMSAT satellite-based estimates; 2015–2017 anomalies and intra-seasonal anomalies during the monitoring period were summarized to contextualize drought conditions. Analysis: Data were cleaned and checked post-download. Functionality, usage, and access indicators were computed weekly by source type and altitude zone, with supplementary analyses by administrative area and altitude bins. Household water quantities (lpcd) were summarized for the final five weeks.

• Functionality: Hand-pumped boreholes had the highest overall functionality (~75%) and motorised boreholes the lowest (~60%) during the monitoring period. In lowland Kolla areas, springs, protected wells and open sources showed large declines; hand-pumps and motorised boreholes initially increased due to intensified operation and maintenance, with hand-pumps maintaining more stable functionality after week 8. In highland Weyna Dega and Dega, functionality increased across all types, especially for motorised boreholes from week 5 onward. Real-time monitoring and rapid, proactive O&M drove gains, particularly for hand-pumps.
• Usage and coverage: Approximately 830,000 people relied on monitored sources (~25% of the population in the monitored areas). Hand-pumped and motorised boreholes together supplied about 400,000 people (~55% at peak). Motorised boreholes served >150,000 people in the Kolla; hand-pumps served >100,000 in the Weyna Dega. About 50,000 people used emergency water trucking, with usage patterns reflecting recovery/failure of shallow sources.
• Demand shifts: In the Kolla, all source types saw significant increases in users per point; median users exceeded 1000 from week 4. Deep groundwater sources (>30 m)—hand-pumps and motorised boreholes—experienced the most regular increases. By week 4, motorised boreholes in the Kolla had median user numbers around 2500, buffering failures of shallow groundwater and surface sources.
• Access (travel time): Travel times increased overall during the drought. In the Kolla, most motorised borehole users traveled >60 minutes; only about 25% of hand-pump users traveled >60 minutes. For springs, travel time decreased after week 8 as rainfall increased and some recovery occurred.
• Quantity sufficiency (perceptions and lpcd): Perceived adequacy of collected water declined over time for most sources. For motorised boreholes, about 50% of users consistently reported sufficient quantities; for hand-pumps, sufficiency fell from ~70% to ~50% by the end. The largest drop (~50 percentage points) occurred for springs in the Kolla. Trucked water users reported the lowest sufficiency. Household surveys in weeks 7–12 showed most users collected <15 lpcd, with the lowest quantities generally in the Weyna Dega. As the drought progressed, Kolla users shifted collection towards motorised boreholes and away from open sources.
• Technology performance: Deep groundwater sources (hand-pumped and motorised boreholes) exhibited high resilience, whereas springs, protected wells, and open sources were vulnerable, particularly in the lowland Kolla. Hand-pumps showed faster repair and sustained functionality compared to motorised systems, which likely suffered longer downtimes due to more complex maintenance and limited access to skills, parts, and fuel.
• Portfolio effect: Access to multiple improved sources and technologies distributed increased demand, reducing reliance on emergency trucking and supporting service continuity.

The study demonstrates that technology choice and groundwater access depth critically influence rural water supply resilience during drought. Hand-pumped and motorised boreholes that tap deep (>30 m) groundwater provided the most reliable services, with hand-pumps achieving higher and more rapidly recoverable functionality likely due to simpler mechanisms and quicker repairs. In contrast, shallow groundwater and surface-dependent sources (springs, protected wells, open sources) were highly vulnerable in lowland Kolla areas, where declines in functionality and increased travel times were most pronounced. The findings support a portfolio approach—maintaining multiple improved water sources and technologies within communities—to spread demand during crises, sustain access, and limit dependency on emergency trucking. Real-time monitoring paired with responsive, externalized O&M improved functionality trajectories, particularly for hand-pumps, underscoring that timely data and clear fault-reporting systems are valuable only when matched with capacity to act. The results also intersect with policy debates: while SDG 6.1 emphasizes safely managed, on-premises services, a strong focus on piped schemes may inadvertently neglect simpler, community point sources that prove resilient during drought and are more feasible in many rural contexts. Strengthening O&M systems, supply chains, and technical skills—especially for motorised systems in deep groundwater settings—is crucial, as is supporting hand-pumped boreholes and reliable springs in highland areas. Overall, the study addresses the research question by evidencing which technologies and management models maintained functionality and access under severe drought stress and by highlighting how operational strategies and diversified infrastructure increase resilience.

Using near real-time, weekly monitoring of thousands of water points during Ethiopia’s 2015–16 drought, the study shows that deep groundwater sources—hand-pumped and motorised boreholes—were most resilient, whereas shallow groundwater and surface sources were vulnerable, especially in lowlands. Rapid increases in functionality during the intervention period were enabled by real-time monitoring and responsive, proactive operation and maintenance, with hand-pumps benefiting most due to simpler repair needs. A community-level portfolio of improved sources distributed demand, limited the need for emergency trucking, and maintained service continuity. Policy and investment should prioritize: (1) strengthening O&M systems and external support mechanisms; (2) investing in technical skills and supply chains for motorised systems in deep groundwater regions; (3) preserving and expanding reliable hand-pumped boreholes and springs where hydrogeologically appropriate; and (4) leveraging smart monitoring for rapid fault detection and response. Future research should assess technical, social, and economic feasibility of piped schemes across varied geographies; quantify life-cycle costs and service reliability for mixed portfolios under different climate scenarios; and evaluate how smart monitoring integrated with service delivery models can sustain high functionality outside of emergency contexts.

Functionality was based on user-reported status (functional/partial/non-functional) with visual confirmation, rather than measured yields or detailed reliability metrics, which may introduce perception bias. Household water quantity data (lpcd) were collected only in the final month, limiting temporal depth for that metric. The intensive, enumerator-driven weekly monitoring over ~16 weeks, while effective, may not be economically or logistically feasible under normal conditions. Motorised borehole repair dynamics may include lag effects not fully captured by weekly snapshots. Findings are contextual to Ethiopia’s 2015–16 drought, specific altitude zones, and sampled Woredas, which may limit generalizability to other settings without similar hydrogeological and institutional contexts. Baseline (pre-intervention) functionality was low, and improvements coincided with proactive O&M interventions, making it difficult to fully disentangle drought impacts from management responses.