Recent Approaches to Robust Water Resources Management under Hydroclimatic Uncertainty

Introduction

Managing water resources under hydroclimatic uncertainty is a primary focus with respect to water resources management (Brown et al. 2015; Poff et al. 2015). There is a scientific consensus that climate change is going to impact current water systems in a variety of ways. For example, expected impacts on California’s water resources include increase in evaporation rates and water demands (Hayhoe et al. 2004); reduction in crop yield as a consequence of higher temperatures (Pathak et al. 2018); less precipitation falling as snow and more as rain translating into wetter wet seasons and drier dry seasons (Mallakpour et al. 2018), which may force operators to lower reservoir levels to allow for space and reduce the risk of flood events; increase in water temperatures affecting aquatic and riparian ecosystem (Poff et al. 2012); and sea-level rise, which will increase seawater intrusion into coastal aquifers (Ferguson and Gleeson 2012). Meanwhile, there is, however, uncertainty on their magnitude and extent, which raise the question of how we can develop management strategies that account for the range of possible future alternatives.

Water systems modeling has been crucial to answer such questions because of the advancement of computational power. Water systems modeling was often deterministically developed, considering some variables such as temperatures, precipitation, streamflow, or water demands as known parameters. Despite recognizing their hydrologic variability, the only available data are often treated as a statistical description of hydrologic variables resulting into unreliable forecasts. Deterministic models treat stochastic parameters as known quantities reducing - to some extent - the complexity of the model (Chen et al. 2018). Depending on the application of the model, such assumptions can be sustained. Nevertheless, the simplifications cannot retain all the essential characteristics of the original data and may lead to unsatisfactory results given the complex behavior of a system (Puente et al. 2018). Climate change is bringing extreme hydroclimatic events of low probabilities but high impacts that - as noted - deterministic approaches do not capture well (Farmer and Vogel 2016; Philbrick and Kitanidis 1999). Thus, there is a need for integrating the seemingly random behavior of precipitation, temperature, streamflow, and other sometimes-unknown variables (e.g., soil properties, water quality) for developing water management strategies suitable for a variety of future socioeconomic and climatic alternatives.

This chapter describes recent approaches from the scientific literature that incorporate hydroclimatic uncertainty to develop robust water management strategies. This chapter is divided into three sections. The first section describes the development of robust reservoir operations, a main technical area in water resources. The second section describes an approach to achieve water systems sustainability that goes beyond common performance criteria (e.g., economic, reliability) by integrating stakeholder-defined performance metrics within the modeling framework. The last section outlines how to move forward and incorporate uncertainty into the decision-making process, a crucial component pertaining to real- life activities.

Robust Reservoirs Management

Reservoirs around the world provide storage to supply water for urban, agriculture, and industry sectors; flood protection; hydropower; and support recreational activities. However, in some cases, these benefits came at the cost of environmental and social degradation (WCD 2000). Leaving aside the controversy of building dams, it is important to recognize that most of the current reservoirs will persist despite their contribution to the degradation of river ecosystems. Given the uncertainty and variability of hydroclimatic conditions under climate change, droughts and floods may worsen, turning reservoir management vital to prevent or reduce expected drought and flood events (Cristina and Tullos 2017).

Modifying reservoir operations offers an opportunity for mitigating hydrologic responses to climate change as current operation rules tend to be static and based on historical inflows and outflows observations. Given the new recurrence and magnitude of hydroclimatic events, some reservoir operation rules are no longer suitable for managing drought periods and floods or for reaching the full potential of a reservoir (Howard 1999; Moy et al. 1986). New reservoir operations that incorporate projected hydroclimatic variability may be an effective strategy for reducing the impacts of changes in water supplies and demands (Vonk et al. 2014). Reservoir operations developed with new available tools and information contribute to increased resilience of water management systems and ecosystem restoration.

Alternative approaches to improve long-term reservoir operations have utilized multi-objective genetic programming (Ashofteh et al. 2015), machine learning (Herman and Steinschneider 2018), and stochastic modeling (Ermoliev et al. 2019) to incorporate climate change projections and a broader set of observations and forecasts. Such approaches link hydroclimatic observations and predictions with water resources management decisions to improve tradeoffs among human and environmental water supply and flood management objectives.

Another recently proposed framework presents a two-stage stochastic optimization model that maximizes regional economic benefits as a function of reservoir deliveries to water users (Ortiz-Partida et al. 2019). The first-stage decisions allocate water based on an expected deficit for the whole system, while in the second stage, the now known deficit is allocated in a way that maximizes economic benefits. This framework was applied to the single reservoir system of Luis L. Leon dam in the Rio Conchos, the main tributary to the Rio Grande of North America. The model result is a set of robust monthly reservoir releases that can cover most of the water demands - including the environment - and reduce the frequency and magnitude of flood events under a wide range of hydroclimatic conditions.

Results from these different approaches suggest that robust operations could improve long-term planning by making the water system more reliable and resilient, and less vulnerable to extreme hydroclimatic events. However, these technically complex approaches tend to have bigger scale implications and usually lack stakeholder involvement, making them difficult to implement.