Robustness-based evaluation of long-term river basin planning under climate change

Abstract:

This work develops a bottom-up, multi-stage planning framework for the sustainable development of river basin systems under deep climate uncertainty. The research focuses on whether and when it is desirable to invest for costly water infrastructure projects, and how to select among a set of project alternatives in order achieve the desired economic benefits with a relatively low level of risk. The proposed framework begins with identifying a set of climate conditions to represent the future vulnerability domain of the system using simulation analysis. The conditions identified in the simulation analysis are then used to develop a scenario-tree, to represent the manner in which the uncertainties may evolve over the course of the planning period. Next, optimal decisions are repeatedly explored through a multi-stage optimization model, by varying the probability weights employed in the scenario-tree. The resulting vector of optimal decisions are post-processed to identify robust choices that are least sensitive to the scenario probabilities.

The proposed planning framework is illustrated for the Niger Basin, over a 45-year planning period from 2015 to 2060. The Niger Basin is a transboundary system facing a series of challenges including endemic poverty, inadequate infrastructure and weak adaptive capacity to climate variability and change. The case study assesses long-term economic benefits from four new dam projects, and from a range of expansions across the eleven irrigation zones. The climate scenarios are obtained by first generating new climate variability realizations from a stochastic weather generator, and then placing climate change factors on the generated climate realizations. Basin runoff response to climate scenarios are simulated by a series of monthly, two-compartment water balance models. Long-term economic benefits are estimated from the sectors of irrigated agriculture, hydropower, navigation, fishing, and environmental protection, using a mixed-integer linear program (MILP).