Redefine Water Infrastructure Adaptation to a Nonstationary Climate

The statement “Climate stationarity is dead” by Milly et al. (2008) stresses the need to evaluate and when necessary, to incorporate nonstationary hydroclimatic changes into water resources and infrastructure planning and engineering. Variations of this theme echo in several other recent editorials by Rogers (2008), Lettenmaier (2008), and Werick and Palmer (2008). The gravity of this topic was firmly felt among participants at a recent USEPA expert and stakeholder workshop on water infrastructure adaptation to climate change (USEPA 2009). Two questions remain at the core: (1) Is climate change “tangible” for consideration in water engineering and planning? and (2) if so, how can we develop “actionable science” for adaptation? In other words, can adaptation practice define the rate of hydroclimatic changes at local scales that are “tangible” and commensurate with other traditional engineering and planning variables? For “tangible” changes, how do we manage the risk arising from hydroclimatic projection uncertainties and justify the adaptation actions to the public and stakeholders?

In this editorial, I will answer a “yes” to the first question and then argue for redefining the adaptation in the context of “how.”

The adaptation need is laid out in many studies that have defined the magnitude and frequency of hydroclimatic changes using climate model simulations or statistical analysis of historical observations or both, although accurate future projections at a local scale remain elusive (IPCC 2007; Bader et al. 2008; Fowler et al. 2007; Mearns et al. 2003). According to the studies described in IPCC (2007) and references therein, global climate change has resulted in substantial warming of atmospheric temperature and a shifting of precipitation regimes in space and time. Such changes will accelerate in this century: a rise of $0.7°\mathrm{C}$ $(1.4°\mathrm{F})$ in temperature over the past $100\text{years}$ as compared to a greater projected increase of more than $4.0°\mathrm{C}$ $(2.4–6.4°\mathrm{C})$ , or $7.2°\mathrm{F}$ $(4.3–11.5°\mathrm{F})$ by 2100 in the worst-case A1F1 emission scenario [Intergovernmental Panel on Climate Change (IPCC) 2007].

The impacts of these changes on water infrastructures and programs are apparent in several ways. Changes in hydrologic fluxes, water quantity, and quality of surface water and groundwater can affect the service functions of a water infrastructure that was designed and built under the assumption of climate stationarity. This compromise can occur in a nonstationary climate where the rate of future hydroclimatic change exceeds the statistical bounds of historical observations. An ongoing EPA study shows the precipitation in the contiguous United States has changed in the past century at a rate comparable to those of population changes; the latter is a traditional variable in the infrastructure planning and engineering. The two principal variables are often mismatched in spatial distribution, resulting in the imbalance of water demand and water availability. The study results also revealed significant spatial variability in the rate of precipitation change among geographic locations, a fact that deserves careful evaluation in planning and designing adaptation at local watershed scales. Because of this, current engineering practices need careful examination: for example, the selection of $25\text{-year}$ return storms, 7Q10 stream flows, base flow and peak runoff for hydraulic works, design basis for combined sewer overflows and storm water management, and wastewater treatment and discharge.

The second driver to redefine adaptation arises from the impending needs to improve our aging water infrastructures. The United States, like other developed countries, has invested substantially in the development and operation of storm water, wastewater, drinking water, and other water infrastructure assets. These physical assets have served us well in providing safe drinking water and wastewater sanitation, the very service functions considered symbolic of civilization and the core of modern-day environmental protection efforts. Because the infrastructure is aging, its service functionality is often vulnerable. A series of USEPA reports (USEPA 2002) detailed the current conditions of water infrastructures and the need for their repair, rehabilitation, and replacement. This urgency was also emphasized by several governmental and nongovernmental entities, including ASCE, who concluded in 2009 that $2.2 trillion is needed over a five-year period to bring the nation’s infrastructure to a good condition, and $255 billion is needed for drinking water and wastewater alone. Aging water infrastructure improvement is also a rare opportunity to proactively incorporate nonstationary hydroclimatic changes and to engineer water infrastructures of greater resilience. Such a need spells out the urgency for actions.

These two principal drivers—tangible hydroclimatic changes and aging water infrastructure improvement—jointly point to the necessity, not just the option, of water infrastructure adaptation in actions ranging from climate change impact assessment, design basis revision, to adaptive engineering. We are witnessing the expansion in adaptation-oriented research activities and an increase in field adaptation applications. There is a growing list of assessment and adaptation tools, methodologies, and guidelines emerging from USEPA, International Water Association, American Water Works Association, ASCE, World Bank, and many other organizations. In this evolving process, we must not overlook the two basic elements of water infrastructure engineering: service functions of an infrastructure to achieve intended management objectives in its lifespan, and the engineering conservatism for adequate resilience against uncertainties at a reasonable cost. Then we can properly define engineering attributes and redefine the adaptation strategies and practices.

Do we embrace adaptation in our traditional practice? Certainly, we do. Planning for future conditions and reserving adequate capacity against uncertainties have been a foundation of water resources planning and infrastructure engineering. The Dujiangyan division and irrigation system on the Ming River in China has experienced climate variations since its construction $\mathrm{2,265}\text{years}$ ago. It still functions today in providing flood protection and drought relief for over $\mathrm{5,300}{\mathrm{km}}^2$ agricultural lands. This example illustrates the potential longevity of a water infrastructure and the varying hydroclimatic conditions under which it has to serve. For modern water infrastructures, forward planning and engineering is the practice to ensure reliable water and wastewater sanitation, and a safe drinking-water supply. Adequate and appropriate capacity reserve is installed for a desired resilience at a reasonable cost. This engineering concept is lectured in college classrooms, practiced in engineering offices, and discussed in management board meetings. To many water resource practitioners, familiar adaptation methods include progressive refining of engineering design basis from conceptual to full design levels, periodic revision of capital improvement master plans, retrofitting of capacity-strained water works, and the application of engineering safety factors, modular design, phased construction, and scenario planning; all are widely used in current engineering practice.

Can we directly apply our familiar engineering techniques in adaptation to hydroclimatic changes? Not necessarily. Future hydroclimatic changes for adaptation are not factual observations, but projections based on the best knowledge of evolving climate and hydrological sciences. These projections are different from the traditional planning and engineering variables due to their levels of uncertainty. For example, the 21 general circulation models in the Program for Climate Model Diagnosis and Intercomparison multimodel database yield precipitation projections of three-month averages in northern America. These projections, when compared with calibrating observation data in control runs, have a maximum bias of 93% in winter season (December, January, and February) for western North America, 21% in the spring season (March, April, and May) for eastern North America, and $-16\mathrm{\%}$ in the fall season (September, October, and November) for central North America (IPCC 2007). As for water infrastructure engineering at a given location, the bias or uncertainty in the $24\mathrm{h}$ precipitation projection is often far greater than the aforementioned levels of bias in regional and quarterly averages. In comparison, progressive revision in traditional engineering practice can reduce the uncertainties in design parameters significantly to the extent to which a simple design safety factor can be used to buffer or protect the system economically and efficiently. This simple, traditional adaptation technique alone is clearly inadequate to manage the degree of uncertainties in hydroclimatic projections.

In closing, the urgency for action on aging water infrastructure and the imperfect climate projections for engineering pose a daunting challenge to be met in the adaptation planning and engineering. This dilemma requires an effective adaptation to strike a balance between projection uncertainties, risk reduction, and adaptation cost. By this token, we need to focus on “tangible” local climate changes and determine their projection uncertainties to the best extent. It is equally important to evaluate and redefine the existing and emerging adaptation techniques (e.g., probability-based safety factor selection, process retrofitting, modular design, low-impact development, and decentralized system applications) for their effectiveness in managing the infrastructure risk from not only the climate change itself, but the uncertainties in its projections.