by Jay Lund
Water supply reliability is a major policy and management goal in California, and in the rest of the world, today and since the beginning of time. The goals of reliable water supplies have grown from supporting human health, to supporting economic prosperity, to supporting healthy ecosystems, even when these goals conflict.
Since ancient times, water supply planning, engineering, and operations have sought to provide reliable water supplies. But until 106 years ago, there was little sophistication on exactly how reliable a water supply would be or should be.
Today, water uses have grown and diversified, and sometimes conflict when water availability is insufficient for all uses. Water availability will always be limited, despite infrastructure investments, and often will diminish or become more expensive with climate change and evolving environmental and public health standards.
However, perfect reliability is never possible, and high reliability often incurs high economic or environmental costs. This has long been the dilemma in California. How do we balance reliability, the costs of improving reliability, and the water shortage costs of unreliability?
Allen Hazen’s 1914 paper established a direction for solving this balancing problem. His paper, “Storage to Be Provided in Impounding Reservoirs for Municipal Water Supply,” assembled streamflow data for more than a dozen water supplies and examined their reliability for a range of water use levels and a range of reservoir sizes. These calculations were done by hand.
Realizing that these streamflow records were short, he further fit these reliability results onto “probability paper” (which he invented) to better estimate reliabilities for different extremes, configurations, and demands.
For water wonks with technical or modeling interests, Hazen’s paper remains well worth reading. Few papers about delivery reliability today thoughtfully synthesize such breadth and detail. Some of the paper’s main lessons remain relevant for water policy and management:
- No water supply can be completely reliable (from an engineer or manager’s perspective).
- Higher reliability requires greater infrastructure costs, with extreme reliabilities incurring extremely large infrastructure and costs.
- Seasonal and over-year water storage should be considered quite differently. Seasonal storage of water in reservoirs requires less storage capacity per unit of reliable delivery. Storing water for over-year droughts requires proportionally more storage capacity and costs. “With a larger reservoir, there is some further gain with increasing size, but in a diminishing ratio.”
- High levels of reliable supply require larger reservoirs, which are costly and will often take years to refill.
- Even an infinite reservoir size cannot reliably deliver more than a reservoir’s average annual inflow.
- Regions with more variable hydrology require greater water storage capacity to supply the same reliability, all else being equal.
- Climate change is likely, but is hard to estimate. There also seem to be longer-term cycles in runoff records, which are difficult to characterize and predict.
- Quantifying water shortage amounts is important. Probability distributions of shortage can be more useful than the mere probability of a shortage. However, probability distributions of shortages are harder to estimate, as shortages are usually rare events.
- Water reliability analysis is inherently approximate, and it is not worthwhile to overly refine data. “In all hydraulic data the probable error of measurement is considerable. There is, therefore, no justification for the application of extreme refinements in methods of calculation.” Evaporation estimates and data are “less adequate than could be desired. Nonetheless, some approximations can be reached.” Longer flow records reduce uncertainty, but do not eliminate it. Even averages have errors. However, probable errors in such estimates can be quantified.
- The natural storage in lakes and sandy stream and lake banks can only be approximated, but can “have a great influence on the required storage, especially at relatively low draft [withdrawal] rates. …”.
- Modeling with monthly flows and operations is somewhat less accurate than daily flows, and tends to under-predict storage needs for a given reliability and other conditions. Weekly time steps correct most of this underestimation.
- Balancing the cost of improving supplies against shortage costs is needed. Reducing water use or adopting other water supplies can be less expensive than expanding reservoirs to increase water storage, especially for infrequent droughts.
- Sometimes hedging reservoir releases, to create more frequent small shortages, can be less damaging overall than accumulating a smaller number of large shortages instead.
- Public displeasure with large drought shortages can lead to infrastructure overinvestment. And the public seeing water spilling from full reservoirs in a few years can encourage the public to think that a supply is not being used to its reasonable limit.
- Hazen extensively discussed limitations of his methods and findings, and estimated and discussed probable errors in his findings. As he summarized, “frank recognition of the large probable errors in many of the results cannot fail to be advantageous.”
Much of today’s work on water supply reliability would advance to reflect some of the methods and thinking from 1914.
Allen Hazen was a founder of modern urban water supply. Many aspects of urban water systems today date back to him and the State of Massachusetts’ Lawrence Experiment Station in the late 1800s. This was where Hazen and colleagues worked on fundamentals of water filtration, later expanding as a consulting engineer to fundamentals of pipe network water distribution, reservoir sizing, water metering, utility finance, and overall integration of urban water systems.
We all drink water (mostly reliably) based on his work. Many of his insights and approaches to water problems remain useful today.
Read old to stay sharp.
Hazen, A. (1914), “Storage to be provided in impounding reservoirs for municipal water supply,” Transactions of the American Society of Civil Engineers, Vol. 77, December, pp. 1542-1669.
Hazen, A. (1909 and 1914), Clean Water and How to Get It, New York, J. Wiley & Sons, 252 pp.
Hirsch, R.M. (1978), Risk Analysis for a Water-Supply System – Occoquan Reservoir, Fairfax and Prince Williams Counties, Virginia, Open File Report 78-452, U.S. Geologic Survey, Reston, VA, also in Hydrologic Science Bulletin, Vol. 23, No. 4, pp. 475-505.
Klemes, V. (1987), “One Hundred Years of Applied Storage Reservoir Theory,” Water Resources Management, Vol 1 , pp. 159-175.