that the project had to begin within 6 months and be completed within 2 years. Yet, he explained that, because the Michigan DOT design pipeline had already been established for $1 billion per year, such a large infusion of funds would result in a dramatic increase in the project pipeline and a significant decrease after the stimulus money was gone. In addition, few projects were on the shelf, and concerns arose about the capacity for private consulting and contractor labor. Therefore, to meet the ARRA criteria, $300 million was allocated for design build, without which the project would not have been delivered on time. He emphasized that time constraints were a significant factor in project selection; political leaders were dissatisfied with the shorter-term projects that were selected, but “transformational projects” were infeasible within the program’s time constraints. Because the goal of the stimulus was to employ as many people as possible, a balance of available work type was also critical; Steudle’s team monitored weekly bids to determine when industry capacity was reached and shifted funds to areas with capacity to employ more people, allow for greater competition, and decrease prices.

To create a plan for success, Steudle reiterated that decision makers should begin with analytics, add a lens to view the reality of the world, and combine the results. Serving as session moderator, Leah Brooks, The George Washington University, asked how environmental concerns affected the Michigan DOT’s decision-making process during its participation in the ARRA stimulus program. Steudle replied that any projects with environmental issues that could not be resolved in time to begin work within 6 months had to be delayed, owing to the time constraints of the program.

1.2 MANAGING UNCERTAINTY IN LOCAL DECISION MAKING

Fischbach underscored that local decision makers face an increasingly complex and uncertain planning landscape, especially as it relates to resilient infrastructure. Often, multiple stakeholders are involved in the decision-making process, each with competing objectives. Other challenges for decision makers revolve around geography and equity (i.e., consideration of outcomes for specific groups so as to target investments accordingly) as well as sustainable financing and implementation. These decision makers would benefit from more data and simulation models to better understand the benefits and costs of potential interventions, which requires increased technical capacity at the local level.

Fischbach explained that decision making under deep uncertainty (DMDU) in local contexts presents a systems challenge. That is, decision makers have to evaluate potential natural and societal outcomes for complex, interrelated, and poorly understood systems; understand

which adaptation responses perform best given limited past experience; and determine how climate change could interact with other uncertainties such as population and land-use shifts. He asserted that traditional risk management methods (in which assumptions about the future are determined first) are less useful in these situations: uncertainties could be underestimated or ignored, competing analyses could contribute to gridlock, and misplaced concreteness could blind decision makers to surprise.

Fischbach noted that for DMDU methods, the analysis is run in reverse: the goal is to reach consensus on the decision by the end of the process without necessarily having to agree on a set of assumptions about the future at the outset of the analysis. This iterative process considers trade-offs across multiple objectives, proposes a strategy, identifies vulnerabilities of the strategy, and develops strategy adaptations to reduce these vulnerabilities. He pointed out that applying DMDU methods for local planning demands significant engagement—by convening several decision makers; building capacity for technical analysis; and developing a range of narratives for planners, stakeholders, and residents.

Using Pittsburgh as an example, Fischbach observed that urbanization, underinvestment, and climate change have overwhelmed the city’s stormwater and wastewater management systems. The Allegheny County Sanitary Authority (ALCOSAN, a regional utility) conveys and treats wastewater from 83 municipalities, most of which have their own wastewater authority, such as the Pittsburgh Water and Sewer Authority. He described a 2017 RAND study that used DMDU methods to identify more robust regional stormwater management strategies. The study convened a participatory, science-based planning process; explored potential water-quality vulnerability to future climate, population, and land-use change; evaluated performance of proposed stormwater/source reduction strategies across a range of uncertainty; and worked with stakeholders to identify improved approaches. The analysis of future vulnerability, which applied models of the ALCOSAN system to simulate sewer overflows across various scenarios, revealed that sewer overflows, already a problem, could exceed current estimates by up to 15 percent. In addition, future rainfall, population, and land-use changes could further increase overflow volumes by 15–40 percent. The study advised that plausible future change, not historical experience, should inform near-term planning and design for stormwater and wastewater infrastructure. A follow-on study was conducted to estimate current and future flooding and overflow risks with no additional infrastructure investment; evaluate sewer overflow and flood-risk reduction from proposed green infrastructure strategies; and compare the water-quality benefits, co-benefits, and costs of proposed green infrastructure strategies. This subsequent study found that 11–18 percent of homes are currently at risk of basement flooding (a

number expected to increase by 18–21 percent in future climate scenarios) and that the net economic value of green infrastructure strategies is nearly always positive across a wide range of scenarios.

Fischbach emphasized that research to inform stormwater planning under uncertainty is ongoing, with new methods for spatial allocation of urban green infrastructure under uncertainty, “dual-drainage” flood modeling, and urban flood analysis under uncertainty at scale.