Vulnerability of Malawi's water, energy and food security to threshold behaviour of Lake Malawi - A lake-basin modelling study

Malawi's energy and food security depends on Lake Malawi outflows into the Shire River, which sustain 98% of hydropower, most of its irrigation, and biodiversity of a Ramsar wetland; the Elephant Marsh. Observations indicate long periods (early 1900s and before 1850s) without lake outflows. Such conditions in the future could adversely affect existing and planned infrastructure. Here, we explore risks due to future climatic and socio-economic uncertainties, and assess robustness of stakeholder-identified adaptation options in addressing risks. Deep uncertainty characterizes these two drivers in the Lake Malawi Shire River Basin (LMSRB) because a) it is affected by East African and Southern African precipitation regimes, and b) land use change and socio-economic development affects water demand. We use the Water Evaluation And Planning (WEAP) model to run scenarios based on stakeholder-identified infrastructural plans, performance needs and adaptation options. We satisfactorily calibrate (1961-1985) and validate (1986-2009) the model for Lake Malawi levels. We analyse a) sensitivity of the lake levels to changes in precipitation using synthetic time series b) lake level projections using a combination of 27 bias-corrected CMIP5 model projections and four socio-economic projections, and c) robustness assessment of adaptation options against performance requirements. We find that the lake is sensitive to precipitation and water demand changes and several future scenarios have periods without outflow. Sectoral performance needs are not met across a large number of scenarios, indicating the extent of infrastructural risk. Individual adaptation options show variable robustness suggesting a) residual risk after adaptation, and b) a combination of adaptation options may be necessary for risk mitigation. Overall, we find value in combining modelling and stakeholder engagement approaches for producing decision-relevant information under uncertainty.