Bureau of Reclamation Banner

Climate Change: Erosion of Water Supply Predictability Under Climate Change?

Project ID: 2726
Principal Investigator: Levi Brekke
Research Topic: Water Operation Models and Decision Support Systems
Priority Area Assignments: 2010 (Climate Change and Variability Research)
Funded Fiscal Years: 2009 and 2010
Keywords: None

Research Question

* Does climate change with warming lead to reductions in water supply predictability?

Reclamation develops seasonal to annual plans of operation based, in part, on forecast water supply. Such forecasts (e.g., of April-July runoff volumes) are critical planning inputs. They start to arrive in January and are updated monthly through spring. In snowmelt-dominated basins, snowpack offers significant predictive information. Under climate change involving warming, the following trend seems possible: precipitation regime of gradually more rain and less snow, leading to reduced snowpack influence on water supply prediction and reduced water supply predictability. The past occurrence, future development, and potential mitigation trends are explored in this study. Findings have ramifications for:

* Interpreting water supply forecasts in the context of contemporary operations planning

* Simulating forecast-based operations in long-term prospective system analyses.

Need and Benefit

Reclamation operators currently develop annual operations scheduled based, in part, on Spring-Summer water supply forecasts issued by federal service providers (e.g., Natural Resource Conservation Service (NRCS) National Water and Climate Center and the National Weather Service (NWS) River Forecast Centers collectively covering the Western 17 states) as well as some state and Reclamation providers (e.g., California Department of Water Resources [CDWR] and the Pacific Northwest [PN] River and Reservoir Operations Group, respectively). Each service provider typically begins issuing April-July volumetric runoff forecasts in January, and provides monthly updates toward the forecast period.

Each forecast provider maintains a set of statistical forecast models specific to basin and issue-month leading up to the April-July and other Spring-Summer forecast periods. The skill of these forecast models stems from:

* Including antecedent precipitation and/or runoff predictors of forecast period runoff, somewhat indicating soil moisture development and current water-year climate

* Including predictors describing basin snow-water content at time-of-issue, which indicates runoff that will eventually happen

Forecast uncertainty is generally associated with weather possibilities from the time-of-issue through the forecast period.

Relative to antecedent precipitation and runoff predictors, snowpack accounts for a considerable share of predictive power. Past climate assessments and contemporary climate projections concur that warming has been occurring in the late-20th century and is expected to continue through the 21st century. This would logically diminish snowpack development during winter, and erode one of the key sources of Spring-Summer water supply prediction.

This raises questions:

* At what rate and to what degree would we expect this predictive power to diminish?

* Can we detect predictability reduction during the late 20th century?

* What are the ramifications of reduced water supply predictability due to past climate change for contemporary operations planning?

* What are the ramifications of reduced water supply predictability under future climate change for long-term systems planning?

* What forecast redevelopment procedures might be recommended to minimize deterioration in predictive power as climate changes?

Currently, Reclamation's forecast service providers are not able to answer these questions. The proposed research addresses these questions by exploring forecast-model maintenance under simulated climate conditions from 1950-2099 in various Western United States basins:

* Simulate hydrologic response to simulated climate, creating simulated basin hydroclimates from 1950-2099

* Implement forecast-model development within these simulated hydroclimates, and maintenance of these models through time

* Evaluate model skill through time as climate changes

Simulated climate data required to support this research are provided by the "Statistically Downscaled WCRP CMIP3 Climate Projections" archive, at . The archive includes 112 downscaled climate simulations over the contiguous United States, with each projection-dataset including: precipitation and temperature, monthly time step from 1950-2099, and 1/8 degree spatial resolution.

Two benefits could be realized from addressing these research questions. Results should provide better understanding on how to interpret contemporary water supply forecasts received from various forecast service providers, understanding that such forecasts are issued in the context of a changing climate. Also, the findings should guide longer-term prospective planning in how simulated forecast-driven operations are represented. For example, in an analysis on a proposed new reservoir serving through about 2070, operational assumptions are required, probably

Contributing Partners

None

Research Products

Contact the Principal Investigator for information about these documents.

This information was last updated on October 30, 2014
Contact the Research and Development Office with questions or comments about this page