Background Floods are the most frequently occurring natural disaster and constitute a significant public health risk. Several operational satellite-based flood detection systems quantify flooding extent, but it is unclear how far the choice of satellite-based flood product affects the findings of epidemiological studies of associated public health risks. Few studies of flooding’s health impacts have used mixed methods to enrich understanding of these impacts. This study therefore aims to evaluate the relationship between two satellite-derived flood products with outpatient attendance and diarrhoeal disease in northern Ghana, identifying plausible reasons for observed relationships via qualitative interviews. Methods A convergent parallel mixed methods design combined an ecological time series with focus group discussions and key informant interviews. Through an ecological time series component, monthly outpatient attendance and diarrhoea case counts from health facilities in two flood-prone districts for 2016–2020 were integrated with monthly flooding map layers classified via the Moderate Resolution Imaging Spectroradiometer (MODIS) and Landsat satellite sensors. The relationship between reported diarrhoea and outpatient attendance with flooding was examined using Poisson regression, controlling for seasonality and facility catchment population. Four focus group discussions with affected community members and four key informant interviews with health professionals explored flooding’s impact on healthcare delivery and access. Results Flooding detected via Landsat better predicted outpatient attendance and diarrhoea than flooding via MODIS. Outpatient attendance significantly reduced as LandSat-derived flood area per facility catchment increased (adjusted Incidence Rate Ratio = 0.78, 95% CI: 0.61–0.99, p < 0.05), whilst reported diarrhoea significantly increased with flood area per facility catchment (adjusted Incidence Rate Ratio = 4.27, 95% CI: 2.74—6.63, p < 0.001). Key informants noted how flooding affected access to health services as patients and health professionals could not reach the health facility and emergency referrals were unable to travel. Conclusions The significant reduction in outpatient attendance during flooding suggests that flooding impairs healthcare delivery. The relationship is sensitive to the choice of satellite-derived flood product, so future studies should consider integrating multiple sources of satellite imagery for more robust exposure assessment. Health teams and communities should plan spatially targeted flood mitigation and health system adaptation strategies that explicitly address population and workforce mobility issues.

Accurate information on flood extent and exposure is critical for disaster management in data-scarce, vulnerable regions, such as Sub-Saharan Africa (SSA). However, uncertainties in flood extent affect flood exposure estimates. This study developed a framework to examine the spatiotemporal pattern of floods and to assess flood exposure through utilization of satellite images, ground-based participatory mapping of flood extent, and socio-economic data. Drawing on a case study in the White Volta basin in Western Africa, our results showed that synergetic use of multi-temporal radar and optical satellite data improved flood mapping accuracy (77% overall agreement compared with participatory mapping outputs), in comparison with existing global flood datasets (43% overall agreement for the moderate-resolution imaging spectroradiometer (MODIS) Near Real-Time (NRT) Global Flood Product). Increases in flood extent were observed according to our classified product, as well as two existing global flood products. Similarly, increased flood exposure was also observed, however its estimation remains highly uncertain and sensitive to the input dataset used. Population exposure varied greatly depending on the population dataset used, while the greatest farmland and infrastructure exposure was estimated using a composite flood map derived from three products, with lower exposure estimated from each flood product individually. The study shows that there is considerable scope to develop an accurate flood mapping system in SSA and thereby improve flood exposure assessment and develop mitigation and intervention plans.

Flooding is the most frequent natural hazard globally, but evidence of its impact on domestic water point contamination remains limited. This study aimed to assess dam-related flooding's impact on microbiological contamination of rural water points and to evaluate agreement of satellite-derived flood maps with ground-based observations of water point flooding. Fieldwork took place in two Ghanaian districts frequently flooded following dam overspill. Fifty-seven water points were tested for bacterial parameters during and immediately after flooding. Forty water points were resampled in the dry season, with the remainder having run dry. Ground-based observations of flooding were compared with three satellite-derived flood maps. Boreholes were less contaminated than wells or surface waters (geometric mean E. coli = 20.2, 175.6, and 590.7 cfu/100 ml, respectively). Among groundwater points, a Wilcoxon signed-rank test indicated significantly greater median E. coli and thermotolerant coliform contamination during flooding (p = 0.025 and p < 0.001, respectively), but Shigella, salmonella, and intestinal enterococci counts were not significantly different between seasons. In contrast, among surface water points, E. coli, Shigella, and Salmonella counts were significantly greater in dry season samples (p < 0.005 for all parameters), possibly reflecting a "concentration" effect. Satellite-derived flood maps had no or low agreement with ground-based observations of water point flooding. Although groundwater quality deteriorated during and after flooding, surface waters were the most microbiologically contaminated in both seasons. The greatest public health risk thus occurred where households switched to surface water collection during or following flood season. Flood risk should be assessed before borehole installation and existing flood-prone boreholes remediated to mitigate population exposure to contaminated water.