In this age of climate uncertainty, water systems of different scale and scope—from large dams to riverine ecosystems to reservoirs and other water-supply systems—are increasingly vulnerable to shifting hydroclimatic conditions. The question of how to mitigate risk and build resilience into these systems gains relevance daily. Global Climate Models (GCMs) have been in development for decades. They continue to evolve with new data and are viewed by many entities as primary sources of insight for future climate conditions.
GCMs generate projections for broad future trends under different CO2 scenarios but are limited in their ability to provide insight for more-localized regions—even with significant post-processing of model outputs. Some have questioned the ability of GCMs to provide practical guidance in the design and development of critical infrastructure. Perceived limitations have brought attention to a different type of assessment that makes use of finer-scaled climate stress tests tailored to the unique parameters and vulnerabilities of a specific water project or system.
A Climate Stress Test from the Bottom Up
Unlike the “top down” approach of a GCM, which assesses system performance using GCM-generated scenarios as the starting point, a climate stress test begins at the “bottom” with a problematic outcome (typically system failure) and builds the test in reverse, identifying the parameters most critical to optimal system performance. The climatic stressors at issue typically are changes in atmospheric variables such as precipitation, temperature, evapotranspiration and others, which can affect the ability of a water system to perform given watershed-level, system-scale and other site-specific factors.
But the nature of change is important, and a different property of a given variable (e.g., mean vs. variance of precipitation) would be considered a distinct stressor in the context of a stress test. The test is designed to determine not just the conditions that lead to failure, but the conditions that could avert a worst-case scenario as well. It’s a sensitivity analysis that determines system outcomes based on a series of plausible changes or combinations of “perturbations.”
Often the conditions tested go well beyond the range presented by GCMs, typically including data points captured by a GCM, but also more. By factoring in multiple types of change, a stress test is capable of simulating interactions relevant to compound risks. So, in theory at least (and increasingly in practice), the test can be an exhaustive exercise that yields more targeted findings and insights into system weaknesses than those obtained with GCM projections alone. Fowler et al (2024) in WIRES Water, “Climate stress testing for water systems: Review and guide for applications” (bit.ly/StressTestingWaterSystems), point to other weaknesses with traditional approaches, suggesting that a system that appears robust under a given GCM scenario may in fact be at risk if hydroclimatic conditions unfold differently or if specific water-system vulnerabilities are inadequately addressed by the GCM model to begin with.
Proponents of climate stress testing believe not only that the methodology can yield robust, climate-resilient policy and management decisions, but that the approach is flexible and inclusive, with the potential to integrate the concerns of different stakeholder groups more effectively than a more-traditional approach. In the case of multiple stakeholders, key stressors would be selected to reflect stakeholder objectives. See Poff et al (2015) for a case study (bit.ly/SustainableWaterMngmt) of the Iowa River that integrates protection of ecological values with a flood-risk assessment of the Coralville Dam and its associated reservoir and levees.
Critical Decisions: Climate “Stressors” and Other Factors
Designing a climate stress test is complex and still experimental, and this column is just a snapshot of what’s involved. Developing a climate test model includes a general assessment of what constitutes system failure and system performance as well as the range of changes and conditions likely to bring about “tipping points” and critical scenarios. This is essentially the phase where a quantitative representation of the system is described: intervention points, system boundaries and critical junctures.
Identification of principal stressors is obviously of primary importance in a climate stress test. Ideally, the goal is to identify the smallest number of climate attributes with significant impact on the system. A limited set of “perturbed attribute values” is developed that represent the set of plausible changes which the system model will be tested against. These correspond to the axes of the scenario-neutral test space. A hydrometeorological and stochastically generated time series is run that simulates the selected perturbations at each critical juncture. System performance under each test scenario allows analysts to assess how the system performs—and which adaptive changes may need to be made going forward, applying “decision scaling” to adjust proposed interventions accordingly.
Culley et al (2021) performed a climate stress risk-assessment on Lake Como reservoir in Northern Italy (bit.ly/CriticalClimateConditions). Critical questions considered were long-term irrigation performance and flood reliability. The study performed a lengthy and structured analysis and used several models to determine which attributes of many were the most likely to provide “information content” for system performance. This analysis discovered that annual average precipitation was the most-important attribute, but other attributes including number of frost days, summer rainfall and average June temperature also were critical. The flood-reliability question was most sensitive to number of frost days and four additional attributes. The study showed that use of specially selected stressors was critical, and results were markedly different than those produced under a limited and more-generic analysis using only default criteria such as temperature and average annual precipitation.
Shoring Up for Uncertainty
In June 2024, the Congressionally appointed committee reviewing Probable Maximum Precipitation (PMP) estimates—standards necessary for high-hazard infrastructure—released a preliminary report (bit.ly/ModernizingPMPs) recommending that PMP estimates no longer represent “an upper bound on rainfall” but rainfall “with an extremely low exceedance probability.” This proposed shift, along with others that rely on model-based probabilistic estimates, aptly define our current era of climatic uncertainty. Going forward, there’s no doubt that GCM projections will remain important for general planning purposes, but climate stress tests offer an additional layer of preemptive review—and our most critical water systems will likely depend on it.
The post Thoughts From Engineers: Climate Adaptation Planning through Stress Testing first appeared on Informed Infrastructure.