Drought risk and adaptation in East and Southern Africa: Capturing human-water feedbacks

East and Southern Africa are highly vulnerable to drought due to its reliance on rain-fed agriculture and livestock herding. These agropastoral systems are sensitive to climate variability and extreme events such as droughts, making proactive adaptation essential for reducing impacts and strengthening resilience. However, drought risk is not only driven by natural rainfall deficits. Human behaviour and interventions can exacerbate or mitigate drought impacts. Understanding how people respond through adaptation measures and how their responses affect drought hazard is therefore critical. This thesis seeks to capture these two-way human–water dynamics in models to better understand drought adaptation in East and Southern African agropastoral contexts. Chapter 2 investigates how farmers in central and southern Malawi use local knowledge to predict drought and integrate it into agricultural decisions. The study explores whether seasonal climate forecasts become more relevant when combined with farmers’ knowledge. A forecast threshold model is built using meteorological indicators observed before the rainy season to predict dry conditions during the season. Local knowledge informs the selection of these indicators. The predictive skill of this model is evaluated and compared to commonly used ENSO-based indicators. The results show that wind and temperature indicators informed by local knowledge better predict the start of the rainy season and dry spells. Incorporating local knowledge can therefore improve forecast accuracy and contextual relevance for farmers. Chapter 3 presents ADOPT-AP, a dynamic drought adaptation modelling framework that combines socio-hydrological and agent-based approaches. It links the spatially explicit Dryland Water Partitioning (DRYP) hydrological model with ADOPT, a behavioural model simulating bounded rational decision-making. The framework captures different spatial human–water interactions. A sensitivity analysis identifies the most influential calibration variables. Irrigation water abstraction emerges as the factor with the strongest effect on both drought hazard and adaptation uptake. Chapters 4 and 5 further develop and calibrate ADOPT-AP to assess policy scenarios. Chapter 4 examines how large-scale commercial farms in upstream areas increase downstream drought risk in the Upper Ewaso Ng’iro catchment in Kenya. It quantifies the impacts of commercial export farming on drought risks and compares them to scenarios in which these farms are replaced by agropastoral communities or forests. The findings show that replacing commercial farms would raise streamflow and groundwater levels. Downstream communities would have shorter water collection distances and slightly higher crop and milk production during drought years. However, these benefits are relatively small compared to the overall severity of drought. Chapter 5 focuses on the 2020–2023 drought in the Horn of Africa and the dynamics of human adaptation during this prolonged event. Two policy scenarios are simulated: increasing community access to extension services and promoting water harvesting. Doubling access to extension services slightly increases adoption of low-cost measures but does not affect uptake of expensive measures, indicating financial constraints as a main barrier. Increasing water harvesting adoption improves milk production and reduces water collection distances. It slightly improves crop production during the first season of the drought but negatively affects the following two seasons. The study also highlights the hydrological trade-offs of water harvesting, as increased upstream water use can reduce downstream streamflow and groundwater recharge. As demonstrated in this PhD dissertation, human–water feedbacks influence drought risks and its impacts on communities. Therefore, these feedbacks must be explicitly considered in drought risk assessments and the design of adaptation policies. The synthesis further highlights practical implications for drought management, addressing aspects related to water, people, land, and governance, all of which are essential for building community drought resilience in East and Southern Africa.