The U.S. Bureau of Reclamation is considering altering its monthly Colorado River forecasting methods in the face of criticism from experts inside and outside the agency that predictions have been too optimistic.
Changing forecast methods could have major ramifications in how the bureau manages the river, water experts say. Larger cutbacks in water deliveries to Arizona, Nevada and California could possibly be triggered, for example.
The agency will consider starting to base its forecasts on the past 20 years of flows into Lake Powell, compared to the 30 years it uses now, a bureau official told the Arizona Daily Star.
Thatβs what several outside scientists have urged the bureau to do, on the grounds that the shorter time period will screen out a lengthy wet period from the 1990s when runoff into the river was much greater than itβs been since 2000.
Such a change would almost certainly make the forecasts more pessimistic than now. Thatβs because the years since 2000 have not only been extremely dry, some scientists have found them to be the driest in the Southwest for the past 1,200 years. Average annual river flows have dropped from about 15 million acre-feet over the 20th century to barely 12 million since 2000.
The bureauβs expressed willingness to consider changes comes after one of its engineers, James Prairie, aired significant reservations about current forecasting methods at a public water conference held in Denver during the summer of 2022.
His concerns closely matched those raised for several years by outside scientists, most prominent among them Brad Udall. Udall, a Colorado State University water researcher, has been at the forefront since the early 2000s in warning about the risks to the Colorado River from climate change. He and several other outside scientists released a report about a year ago that criticized the accuracy of bureau forecasts.
Predicting availability of water
At issue are whatβs called β24-month studiesβ that the bureau releases every month. The studies use computer models to forecast reservoir water levels for each of the upcoming 24 months. These monthly forecasts are well publicized and regularly discussed in the media by a wide range of water experts, including scholars, government water officials, activists, and engineers and hydrologists in the private sector.
A change to a 20-year forecasting model could have major implications for how the bureau plans its annual releases from Lake Powell to Lake Mead and how it determines the levels of future shortages in river water deliveries from Mead to Lower Basin states, said another outside critic, Eric Kuhn. Heβs an author and researcher and retired general manager of the Colorado River Water Conservation District in Glenwood Springs, Colorado.
For one, it could affect the scale of cutbacks that the bureau makes annually from Lake Mead into the riverβs Lower Basin of Arizona, Nevada and California. Thatβs because the bureau bases the amount of cutbacks, if any, it will make for the following year on what its monthly forecast predicts in August will be the most likely water level at Lake Mead by the end of the current year.
It could also affect how much water the bureau releases each year from Lake Powell to Mead β an amount also crucial for river water availability for Arizona, California and Nevada, Kuhn said.
βThese forecasts are the best eyes we have on the future,β Udall said. βIf they are grossly off, as they have been for many years, we will not make the best possible decisionsβ about managing the river.
The biggest change that will occur if the method changes will be in the bureauβs prediction of the lowest likely river flows β officially called βminimum probableβ flows β expected over an upcoming year or longer, Udall said.
βRecent minimum probable forecasts have utterly downplayed our risk, as we have unfortunately found out recently,β when the bureau suddenly announced last June that drastic cuts need to be made by 2023 in river basin water use, Udall said.
Information βmay be biasedβ
The bureauβs Prairie raised similar concerns at a public conference in August 2022, sponsored by the advocacy group Colorado Water Congress.
βI do not have a lot of confidenceβ in federal forecasts of reservoir levels looking about a year ahead, Prairie said. He cited what he said are biases that are built into the forecast methods because of their reliance on temperatures and precipitation from the 1990s.
βIβm going to want to tend to skew that down,β Prairie said.
Prairie is chief of the bureauβs Colorado Basin Research and Modeling Group and has been a hydrologic engineer for the agency since 2000. He has extensive experience directing and coordinating research, development and modeling projects. He leads applied research in a wide range of topics, including long-term water resource planning and climate variability.
βAs you look at this, I would not walk away thinking this is exactly where we are going to end up,β Prairie said at the conference, referring to the bureauβs forecast for the most likely amount of flows into Powell in 2023. βI would likely lean toward these lower flows, and consider that as youβre thinking about how to protect the system more.β
He told the gathering that when he works with βfolks,β he will tell them that the information they are looking at from the bureau βmay be biasedβ because of the agencyβs use of 30 years of data.
βThat 30-year period is a standard of NOAA (the National Oceanic and Atmospheric Administration). Thatβs a standard view of climate. The way they look at the concept of climate, that idea is not something thatβs going to change quickly,β Prairie said. βItβs something weβve been talking about, but it is embedded at the international level.
βThatβs one reason it is going to be hard for us to move away from those consistent methods, but itβs something to think about when you look at the results,β Prairie said.
A βwet biasβ
The bureau declined to comment on Prairieβs remarks. But in an email to the Star, a Reclamation official said the 30-year period traditionally used for forecasting could be trimmed to 20 years sometime before authorities approve new, broader guidelines for operating the riverβs reservoirs after existing guidelines expire in 2026.
The official acknowledged the current forecasting system has a βwet bias,β but added that there would be tradeoffs if the forecasts were to reflect 20 years of flows instead of 30.
The official noted the Colorado River Basin βhas been seeing an increased temperature trend in the past 30 years. We also see the observed record observes a wetter period from 1990-1999 than we have experienced during the current drought beginning in 2000.
βGiven the warming trend, we believe excluding the early wet years (1990-1999) could reduce the wet bias we are seeing in the recent forecasts,β said the official, who declined to be quoted by name.
βSpread represents the range of forecast β if itβs too tight, you could miss the observation (too high or below it),β said the official, in explaining tradeoffs from switching forecasting methods. βIf itβs too wide, then you donβt capture the specific forecast. Trying to balance the trend between accuracy and spreadβ is the need.
βUsing a shorter forecast period may result in an overall drier forecast but may also result in narrower range of runoff possibilities (e.g., any future wet periods may not be forecast accurately),β Ashley Nielson, a senior hydrologist for the federal Colorado Basin River Forecast Center, has told the Star. βTo account for the large amount of hydroclimactic variability within the Colorado River Basin, we cannot only rely on dry scenarios.β
The possibility that forecasting methods could be tightened comes a little more than a year since the bureau last altered them. Until then, it relied on river flows from 1981 to 2010, but has since changed that period to 1991 to 2020. The change being considered would make that 2000 to 2020.
Critics say a 30-year period isnβt workable in a rapidly changing climate.
βWhen I teach classes, you always teach that using more data is better,β said Professor Jack Schmidt, a longtime Utah State University water researcher. βUnfortunately, in a changing, warming world, having a long averaging period isnβt necessarily better.β
Udall, Kuhn and Schmidt collaborated with others on a February 2022 study by Utah Stateβs Center for Colorado River Studies. It concluded the Bureau of Reclamationβs 24-month studies issued from 2010 through 2021 made projections for the most likely flows that in some cases were 7 million acre-feet a year more than what actually flowed into Powell.
Itβs widely accepted among many scientists that using 30 years isnβt appropriate in an era of climate change, Udall said, βbut the problem is that there is no new standard. It leaves everybody searching for a consensus (forecasting) process that doesnβt exist.β
Photos: The receding waters of Lake Powell, Glen Canyon National Recreation Area