Climate Change Feature

Better Subseasonal-to-Seasonal Forecasts for Water Management

Emerging methods that improve precipitation forecasting over weeks to months could support more informed resource management and increase lead times for responding to droughts and floods.

By , F. Martin Ralph, Duane E. Waliser, Jeanine Jones, and Michael L. Anderson

California experiences the largest year-to-year swings in wintertime precipitation (relative to its average conditions) in the United States, along with considerable swings within a given water year (1 October to 30 September). For example, 1977 was one of the driest years on record, whereas 1978 was one of the wettest. In December 2012, California was on pace for its wettest year on record, but starting in January 2013, the next 14 months were drier than any period of the entire 100-year observational record.

The considerable variability of precipitation within given water years and from year to year poses a major challenge to providing skillful long-range precipitation forecasts. This challenge, coupled with precipitation extremes at both ends of the spectrum—extremes that are projected to increase across the state through the 21st century as a result of climate change—greatly complicates smart management of water resources, upon which tens of millions of residents rely.

The predictive skill of long-range precipitation forecasts in this region has historically been weak, meaning scientists have not been able to aid state and local water managers with reliable forecasts of precipitation and drought for lead times longer than a week or two. The marginal success that forecasters have had to date in predicting winter season rainfall deviations, or anomalies, in California has been tied to the state of the El Niño–Southern Oscillation (ENSO). Yet ENSO explains only a fraction of the historical year-to-year variation in precipitation over California [e.g., DeFlorio et al., 2013], and many predictability studies have been limited by insufficient observational data that only recorded several large ENSO events. Further limitations have been imposed by climate models that did not have accurate enough representations of ENSO and its associated impacts on California’s weather and climate.

These weaknesses have hindered long-range planning and sometimes resulted in reactive or less-than-optimal management decisions. Now, however, California and other states stand to benefit in many ways from emerging research methods that have the potential to improve the skill of subseasonal (2- to 6-week) to seasonal (2- to 6-month) precipitation forecasts. Such forecasts could help, for example, in managing state water supplies during winters with periods of prolonged drought. Long-lasting drought conditions present unique challenges, such as the necessity for drought response activation at the state level.

A cow stands on bare ground near a dry watering hole.
A cow stands near a dry watering hole on a California ranch during drought conditions in 2014. Improved subseasonal-to-seasonal weather forecasts could benefit agriculture and ranching, among other sectors. Credit: U.S. Department of Agriculture photo by Cynthia Mendoza, CC BY 2.0

Responding to the substantial demand from end users, including water managers, the international research community has been increasingly focused in recent years on improving forecast skill and quantifying forecast uncertainty on subseasonal-to-seasonal (S2S) timescales [National Academies of Sciences, Engineering, and Medicine, 2016; Vitart et al., 2017]. Several collaborative efforts within the applied research community have detailed the potential value of S2S forecasts to a variety of end users, including (but not limited to) water resource management. Additional end user sectors that stand to benefit from improved S2S forecasts include agriculture, insurance and reinsurance, and commodities trading [Mariotti et al., 2020; Merryfield et al., 2020].

Stakeholder Needs Drive Investments in S2S Forecasting

Worldwide, the focus on S2S forecasting is steadily increasing. This impetus is represented in the World Meteorological Organization’s World Weather Research Programme (WWRP) and the S2S Prediction Project under the World Climate Research Programme (WCRP). Nationally, the U.S. Weather Research and Forecasting Innovation Act of 2017 (Public Law 115-25) mandated that NOAA improve S2S forecasts to benefit society.

Accordingly, NOAA’s Modeling, Analysis, Predictions and Projections (MAPP) program has led the development of the Subseasonal Experiment (SubX) over the past several years. This effort aims to improve subseasonal prediction of precipitation and other climate variables and to provide a public data set for the research community to explore in predictability studies [Pegion et al., 2019].

Separately, since 2017, the California Department of Water Resources (CDWR) has funded a partnership to improve S2S prediction of precipitation over the western United States, with a particular focus on California. This partnership includes the Center for Western Weather and Water Extremes (CW3E), the NASA Jet Propulsion Laboratory (JPL), and other institutional collaborators. CDWR’s motivation is largely to support drought preparedness—as long ago as California’s 1976–1977 drought, state water managers recognized that the skill of available operational seasonal precipitation forecasts was insufficient for decisionmaking.

The objective of the CW3E-JPL partnership is to provide water resource managers in the western United States with new experimental tools for S2S precipitation forecasting. One such tool, for example, addresses atmospheric rivers [Ralph et al., 2018], or ARs (e.g., the Pineapple Express, one “flavor” of AR), and ridging events (elongated areas of high atmospheric pressure) [Gibson et al., 2020a], both of which strongly affect wintertime precipitation over the western United States [e.g., Guan et al., 2013; Ralph et al., 2019].

The efforts of the CW3E-JPL team are also a part of the S2S Prediction Project’s Real-Time Pilot Initiative. This initiative includes 16 international research groups, each of which is using real-time forecast data from particular modeling centers, along with the S2S Prediction Project’s hindcast database for applied research efforts with a specific end user. Examples of end users participating in this project include the Kenya National Drought Management Authority, the Italian Civil Protection Department, and the Agriculture and Food Stakeholder Network of Australia’s Commonwealth Scientific and Industrial Research Organisation.

The S2S research and development effort described here is the only project in the pilot initiative that is focused on water in the western United States, and it is helping raise the visibility of the needs of western U.S. water resource managers among the international applied science community.

Different Decisions Require Different Lead Times

Water management in California and across the western United States is a challenging and dynamic operation. In addition to the fundamental influence of rainfall and snowfall in determining water supply, water management is affected by many political and socioeconomic considerations. Such considerations in water management include public health and safety minimum supply requirements for the population, which are particularly relevant during extreme drought conditions. Another consideration relevant during less extreme drought times is the prioritization of water use when there is an insufficient amount of resources to meet all objectives (balancing use for fisheries, agriculture, municipalities, etc.).

Effective management of water supply across the region requires different information at different lead times, in part because a variety of atmospheric and oceanic phenomena influence precipitation over these different timescales (Figure 1).

Figure indicating timescales of water management decision support need lead times and physical processes affecting precipitation predictability
Fig. 1. Lead times for water management decision support needs vary over daily to decadal/century timescales, as do physical processes that affect the predictability of precipitation over the western United States.

Weather information provided over shorter lead times provides intelligence for operational decisions regarding flood risk management, emergency response, and situational awareness of potential hazards. Precipitation anomalies on the timescales of weather across the western United States are dominated by the presence or absence of ARs and ridging events. ARs are associated with bringing precipitation to the western United States. They can be beneficial or hazardous from a water management perspective, depending on AR intensity, duration, and antecedent drought conditions [Ralph et al., 2019]. Ridging events are areas of extensive high atmospheric pressure anomalies in the midtroposphere. Several different ridge types have been historically linked to drought over California [Gibson et al., 2020a].

Forecasts with lead times of weeks or months are more useful for decisions about asset positioning or about operational plans that can be adapted to weather outcomes as they happen. For example, state regulations associated with the California State Water Project limit water transfer amounts across the Sacramento–San Joaquin Delta. These water transfers occur because most of California’s water supply originates north of the delta, while most of the demand is south of the delta. Development of water resources infrastructure over the past century has made use of natural waterways to move water from the supply-rich region to the demand centers. Regulatory limits on water transfer could be better supported if we had improved precipitation forecasts with a lead time of weeks to months.

In addition, hydropower systems that have a chain of reservoirs could leverage better S2S forecasts to maximize the value gained from knowing which reservoirs are at capacity and which are running low at any given time. Precipitation anomalies on these timescales are influenced by both ARs and ridging, as well as by variations in the magnitude and phase of ENSO and the Madden–Julian Oscillation, a tropical atmospheric disturbance that travels around the planet every 1–2 months, for example.

On seasonal to annual scales, forecasts aid decisionmaking with respect to resourcing and budgeting that allow water managers to be prepared to respond to weather extremes, or to adopt more costly response packages that may involve legal review components such as environmental review or concurrence with regulatory mandates. Precipitation anomalies at these lead times can be influenced by ENSO and the quasi-biennial oscillation, a quasiperiodic oscillation of equatorial zonal wind anomalies in the stratosphere.

Beyond those scales, longer-term projections of climate change are used for planning adaptation and mitigation strategies. Identifying change thresholds in average precipitation or precipitation extremes can be used as triggers for implementing these strategies, which may require negotiated legislation or longer-term investment strategies.

A key goal of CDWR’s investment in near-term experimental forecasting products is to catalyze improvements in precipitation forecasting to fully implement the S2S requirement of Public Law 115-25. The need for such improvements was highlighted in the National Weather Service’s first-ever service assessment for drought, which summarized California’s drought in 2014 and stated, “A majority of the stakeholders interviewed for this assessment noted one of the best services NOAA could provide is improved seasonal predictions with increased confidence and better interpretation.”

Emerging Technologies Provide New Capabilities

In response to the substantial need in the western U.S. water management community for better S2S precipitation forecasts, CW3E and JPL have developed a suite of research projects using emerging technical methods (Figure 2). For example, deep neural network techniques in combination with large ensemble climate model simulations will support the creation of experimental S2S forecast products for the region.

These products combine both dynamical model output from the S2S database and novel statistical techniques, including machine learning methods applied to large ensemble data sets and mathematical methods for discovering patterns and associations, such as extended empirical orthogonal function analysis and canonical correlation analysis. The experimental forecast tools are supported by peer-reviewed hindcast assessments, which test the skill of a model by having it “predict” known events in the past. There is a particular focus on applying these emerging methods to longer lead times, ranging from 1 to 6 months, over the broad western U.S. region.

Quantities, methods, and lead times investigated to benefit water management in the western United States
Fig. 2. Quantities of interest, methods, and lead times investigated by the Center for Western Weather and Water Extremes/Jet Propulsion Laboratory S2S team to benefit water management in the western United States.

Critically, stakeholders at CDWR involved in water supply forecasting, reservoir operations, and interactions with governance for drought response provide not only funding but also direct input on the design of both research methodologies and the accompanying experimental forecast products. This research and operations partnership exemplifies an efficient applied research pipeline: End users of the forecast products ensure that the research supporting the products is designed and implemented in ways that will be useful to meet their needs, while at the same time, scientific peer review assures these end users of the forecasts’ scientific rigor.

Recently, this partnership has yielded two primary new products that are now available online and are focused on forecasting the odds of wet or dry conditions in coming weeks across the western United States. Each of these methods has been described in detail in formal publications that include quantification of their skill and reliability [DeFlorio et al., 2019a, 2019b; Gibson et al., 2020a, 2020b].

As weather across California and the U.S. West becomes increasingly variable and more difficult to prepare for, new science-based research and operations partnerships like these and others (e.g., Forecast Informed Reservoir Operations, which has supported better water supply management through skillful short-range forecasts of ARs and precipitation [Jasperse et al., 2020]) are offering enhanced abilities to see weeks and months into the future, a vital benefit for water management across the region.

References

DeFlorio, M. J., et al. (2013), Western U.S. extreme precipitation events and their relation to ENSO and PDO in CCSM4, J. Clim., 26(12), 4,231–4,243, https://doi.org/10.1175/JCLI-D-12-00257.1.

DeFlorio, M. J., et al. (2019a), Global evaluation of atmospheric river subseasonal prediction skill, Clim. Dyn., 52, 3,039–3,060, https://doi.org/10.1007/s00382-018-4309-x.

DeFlorio, M. J., et al. (2019b), Experimental subseasonal-to-seasonal (S2S) forecasting of atmospheric rivers over the western United States, J. Geophys. Res. Atmos., 124(21), 11,242–11,265, https://doi.org/10.1029/2019JD031200.

Gibson, P. B., et al. (2020a), Ridging associated with drought across the western and southwestern United States: Characteristics, trends, and predictability sources, J. Clim., 33(7), 2,485–2,508, https://doi.org/10.1175/JCLI-D-19-0439.1.

Gibson, P. B., et al. (2020b), Subseasonal-to-seasonal hindcast skill assessment of ridging events related to drought over the western United States, J. Geophys. Res. Atmos., 125(22), e2020JD033655, https://doi.org/10.1029/2020JD033655.

Guan, B., et al. (2013), The 2010/2011 snow season in California’s Sierra Nevada: Role of atmospheric rivers and modes of large-scale variability, Water Resour. Res., 49(10), 6,731–6,743, https://doi.org/10.1002/wrcr.20537.

Jasperse, J., et al. (2020), Lake Mendocino forecast informed reservoir operations: Final viability assessment, Univ. of Calif., San Diego, https://escholarship.org/uc/item/3b63q04n.

Mariotti, A., et al. (2020), Windows of opportunity for skillful forecasts subseasonal to seasonal and beyond, Bull. Am. Meteorol. Soc., 101(5), E608–E625, https://doi.org/10.1175/BAMS-D-18-0326.1.

Merryfield, W. J., et al. (2020), Current and emerging developments in subseasonal to decadal prediction, Bull. Am. Meteorol. Soc., 101(6), E869–E896, https://doi.org/10.1175/BAMS-D-19-0037.1.

National Academies of Sciences, Engineering, and Medicine (2016), Next Generation Earth System Prediction: Strategies for Subseasonal to Seasonal Forecasts, 350 pp., Natl. Acad. Press, Washington, D.C., https://doi.org/10.17226/21873.

Pegion, K., et al. (2019), The Subseasonal Experiment (SubX): A multimodel subseasonal prediction experiment, Bull. Am. Meteorol. Soc., 100(10), 2,043–2,060, https://doi.org/10.1175/BAMS-D-18-0270.1.

Ralph, F. M., et al. (2018), Defining “atmospheric river”: How the Glossary of Meteorology helped resolve a debate, Bull. Am. Meteorol. Soc., 99, 837–389, https://doi.org/10.1175/BAMS-D-17-0157.1.

Ralph, F. M., et al. (2019), A scale to characterize the strength and impacts of atmospheric rivers, Bull. Am. Meteorol. Soc., 100(2), 269–289, https://doi.org/10.1175/BAMS-D-18-0023.1.

Vitart, F., et al. (2017), The Subseasonal to Seasonal (S2S) Prediction Project database, Bull. Am. Meteorol. Soc., 98(1), 163–173, https://doi.org/10.1175/BAMS-D-16-0017.1.

Author Information

Michael J. DeFlorio ([email protected]) and F. Martin Ralph, Center for Western Weather and Water Extremes, Scripps Institution of Oceanography, University of California, San Diego, La Jolla; Duane E. Waliser, NASA Jet Propulsion Laboratory, California Institute of Technology, Pasadena; also at Joint Institute for Regional Earth System Science and Engineering, University of California, Los Angeles; and Jeanine Jones and Michael L. Anderson, California Department of Water Resources, Sacramento

Citation: DeFlorio, M. J., F. M. Ralph, D. E. Waliser, J. Jones, and M. L. Anderson (2021), Better subseasonal-to-seasonal forecasts for water management, Eos, 102, https://doi.org/10.1029/2021EO159749. Published on 23 June 2021.
Text © 2021. The authors. CC BY-NC-ND 3.0
Except where otherwise noted, images are subject to copyright. Any reuse without express permission from the copyright owner is prohibited.