Water Resources Planning Under (Deep) Uncertainty

Public investments in water infrastructure continue to grow where developed countries prioritize investments in operation and maintenance while developing countries focus on infrastructure expansion. The returns from these investments are contingent on carefully assessed designs and operating strategies that consider the complexities inherent in water management problems. These complexities arise due to several factors, including, but not limited to, the presence of multiple stakeholders with potentially conflicting preferences, lack of knowledge about appropriate systems models or parameterizations, and large uncertainties regarding the evolution of future conditions that will confront these projects. The water resources planning literature has therefore developed a variety of approaches for a quantitative treatment of planning problems. Beginning in the mid-20th century, quantitative design evaluations were based on a stochastic treatment of uncertainty using probability distributions to determine expected costs or risk of failure. Several simulation–optimization frameworks were developed to identify optimal designs with techniques such as linear programming, dynamic programming, stochastic dynamic programming, and evolutionary algorithms. Uncertainty was incorporated within existing frameworks using probability theory, using fuzzy theory to represent ambiguity, or via scenario analysis to represent discrete possibilities for the future. As the effects of climate change became palpable and rapid socioeconomic transformations emerged as the norm, it became evident that existing techniques were not likely to yield reliable designs. The conditions under which an optimal design is developed and tested may differ significantly from those that it will face during its lifetime. These uncertainties, wherein the analyst cannot identify the distributional forms of parameters or the models and forcing variables, are termed “deep uncertainties.” The concept of “robustness” was introduced around the 1980s to identify designs that trade off optimality with reduced sensitivity to such assumptions. However, it was not until the 21st century that robustness analysis became mainstream in water resource planning literature and robustness definitions were expanded to include preferences of multiple actors and sectors as well as their risk attitudes. Decision analytical frameworks that focused on robustness evaluations included robust decision-making, decision scaling, multi-objective robust decision-making, info-gap theory, and so forth. A complementary set of approaches focused on dynamic planning that allowed designs to respond to new information over time. Examples included adaptive policymaking, dynamic adaptive policy pathways, and engineering options analysis, among others. These novel frameworks provide a posteriori decision support to planners aiding in the design of water resources projects under deep uncertainties.