Integrating Quantitative, Ensemble Forecasts into Water Utility Planning and Operations

Water resources agencies and utility managers across the country face daily, complex challenges in reliably supplying the public with water to support domestic use, irrigation, and production. Additionally, water utilities and resource agencies are increasingly relied upon to be stewards of natural waters, requiring increased consideration of recreational, environmental, and flood mitigation objectives. Meanwhile, risks to supply reliability and water quality due to changing climate, economic uncertainties, changes in population dynamics, and other factors threaten to upset best-laid plans for continuing to provide safe and reliable water well into the future.

In many ways, it is because of the wider ecological footprint of water resources that uncertainties in climate and other physical and social systems become so important in water resources planning. To address these challenges, water utility planners and operations staff are increasingly seeking approaches to better quantify risk in alternative long-term planning scenarios as well as in near-term operations decisions. Among the many uncertainties facing water resources managers, some can be represented mathematically, assigned to model inputs or variables, and thereby explicitly incorporated into a water supply planning or operational decision-making framework.

Three common sources of uncertainty and risk in water supply planning and operations are future hydrology, demand, and water quality, all of which are linked to climate, weather, and hydrology dynamics. Traditional planning approaches typically address these uncertainties through scenario analysis, in which “average day” or “stress” conditions are used to evaluate performance of alternatives. This is fundamentally a deterministic analysis, in which the future is represented by a single set of predictions, often based on historical data. Given climate non-stationarity in addition to natural variability, historical conditions are exceedingly unlikely to occur in the same sequence in the future. Further, this approach does not offer a way to establish the risks of adverse or unsatisfactory outcomes and develop mitigation actions, a necessary component of a robust water supply plan.

In contrast, probabilistic supply and demand forecasts are a way of addressing reliability and resilience in performance evaluations by better representing the range and likelihood of possible system responses to uncertain future conditions. Probabilistic or stochastic forecasts can vary in complexity, from simple statistical approaches to more advanced physically-based models or combinations of the two. Hydrologic forecasts can be based on meteorological output from climate models, while demand forecasts can be based on economic, population, and meteorological projections. Similarly, it may be beneficial for utilities to use a stochastic modeling approach to quantify the uncertainty/risk of unacceptable future treatment costs due to uncertainties in water quality conditions. For example, rather than simulate water quality under a single set of conditions (e.g., a design storm), utilities could use a range of input scenarios to simulate water quality under a wide range of hydrologic conditions, potentially tied to probabilistic or stochastic hydrologic inputs.

While the analytical capabilities of the water utility community vary widely, a number of utilities and resource agencies have begun to invest in tools to obtain, process, and analyze quantitative, ensemble forecasts for both long-term planning and near-term decision-making. As part of the panel “Decision Making by Water Utilities: Using Climate/Weather Information in Short and Long Term Planning,” Dr. Weiss will draw from his extensive experience working with a variety of water utilities and resource agencies to integrate ensemble forecasts into long-term planning and near-term operations. Two key examples are his work on developing and implementing the New York City Department of Environmental Protection’s (NYC DEP) Operations Support Tool (OST); and a Drought Planning Tool for the Susquehanna River Basin (SRB DPT).

OST is a linked water supply-water quality decision support tool used by NYC DEP for a variety of planning needs as well as to guide real-time operations. Originally developed as a long-term planning tool to support evaluation of alternative turbidity control approaches for DEP’s unfiltered water supply system, OST has grown into a real-time decision support system. Initialized with real-time information on current reservoir levels and aqueduct settings and water quality status; and driven by the National Weather Service’s Hydrologic Ensemble Forecasting System (HEFS) forecast ensembles, OST is used by NYC DEP operations staff on a daily basis to inform operating decisions to balance a complex set of water supply, water quality, regulatory, environmental, and other objectives. In this panel, Dr. Weiss will describe how OST and HEFS ensemble forecasts have greatly improved decision-making for the NYC water supply, with examples including: maintenance of DEP’s EPA Filtration Avoidance Determination; dynamic operation of Delaware River Basin reservoirs to better balance NYC’s water supply needs with lower Delaware Basin uses and objectives; and planning for the 6-9 month shutdown and repair of the Delaware Aqueduct, which typically supplies 60% of NYC’s drinking water.

In partnership with the Susquehanna River Basin Commission, Dr. Weiss recently led the development and application of a system model-based Drought Planning Tool for the Susquehanna River Basin (SRB DPT). The SRB DPT incorporates a variety of hydrologic and meteorological drought early warning indices (e.g. PDSI, PHDI, SPI) and forecasts into a system model of the Basin to enable development and testing of index- and forecast-based operating strategies for Basin stakeholders. Case studies for the City of Baltimore (MD), Capital Region Water (Harrisburg, PA), and York Water (York, PA) demonstrated the potential for modified operations in the early stages of a drought can provide improved system reliability in later stages. Future upgrades to the DPT and connection with real-time information, including planned HEFS forecasts for the Basin, will enhance stakeholders’ ability to use it to implement proactive drought operations in the near term.

Dr. Weiss will draw on these and other examples to provide a perspective of how water utilities are currently using forecasts and other climate/weather information to make decisions, and how these decisions are improved by quantification of these critical uncertainties.