Seasonal streamflow forecasts

Overview

Seasonal streamflow forecasts, also called seasonal water supply forecasts, are widely consulted by water managers and water users throughout the Colorado River Basin to inform general expectations for the coming runoff season and to inform specific operational decisions such as reservoir release schedules and water allocations. Forecasts are typically issued from mid-winter (Dec/Jan) through July for a several-month forecast period, April-July being the most common, and in some cases for the water year (Oct-Sep). The forecasts are expressed in terms of streamflow volume (in acre-feet, af) over the forecast period.

There are two primary federal providers of operational seasonal streamflow forecasts for the basin:

• Colorado Basin River Forecast Center (CBRFC), one of the 13 NOAA National Weather Service (NWS) River Forecast Centers (RFCs)
• Natural Resource Conservation Service (NRCS) National Water and Climate Center

CBRFC and NRCS use different modeling approaches and their respective forecasted streamflow volumes for the same forecast point may differ. Before 2012, they coordinated their forecasts so that their forecasted streamflow volumes were the same. CBRFC issues forecasts for 143 points within the Colorado River Basin, and also for 33 points in the eastern Great Basin. NRCS issues forecasts for 134 points within the basin, most of them overlapping with the CBRFC forecast points, and also for about 750 points in other western river basins.

Figure 1. CBRFC forecast evolution plot showing the multiple probabilistic forecasts of April-July Lake Powell inflows that were made at each stage of the 2020 forecast season as the snowpack evolved. Image: CBRFC; Annotations: Jeff Lukas.

While users typically focus on the "most probable" (50% exceedance) forecasted values, the operational forecasts from both CBRFC and NRCS include multiple values for each forecast point, to represent the uncertainty in the forecast (Figure 1). These are typically expressed in terms of probability-of-exceedance, in which the 90% exceedance is a low-end forecast, with a 90% probability of being exceeded by the actual streamflow volume, and the 10% exceedance is a high-end forecast, with only a 10% probability of being exceeded by the actual streamflow volume. NRCS has a guide to interpreting streamflow forecasts that explains how users can align their own risk tolerance with a particular exceedance forecast.

A number of basin water agencies produce their own seasonal streamflow forecasts for gages not covered by CBRFC or NRCS, typically using statistical forecast methods similar to those used by NRCS (see below).

Relevance

CBRFC seasonal streamflow forecasts are key inputs into Reclamation's two primary operational models for the basin, 24-Month Study and MTOM, and strongly influence operational decisions and planning by Reclamation. Many other water agencies and water users across the basin consult CBRFC and/or NRCS seasonal streamflow forecasts for their own modeling and decision-making.

Methods

In the snowmelt-dominated headwaters of the Colorado River Basin, predictability of streamflow on seasonal timescales in the basin arises from the watershed's initial moisture conditions: the observed snowpack and soil moisture conditions. The skill of the forecasts increases through the winter and spring months as the accumulating snowpack provides more information about the upcoming runoff. Soil moisture deficits (or surpluses) in the fall, before snow begins to accumulate, indicate whether an an unusually low (or high) proportion of the snowpack will become runoff in the stream.

Lower-elevation watersheds which have little or no snowmelt contributing to the annual hydrology cannot be forecast as skillfully as higher-elevation watersheds that are snowmelt-dominated. In all cases, the primary source of forecast uncertainty and error is due to lack of knowledge of how the weather and climate will evolve between the date the forecast is issued (e.g., April 1st) and the end of the forecast period (e.g., July 31st). Other uncertainty and error comes from inaccurate depictions of the snowpack and/or soil moisture as input into the forecast model, and biases in the model itself.

Figure 2. Example of CBRFC ESP forecast showing 35 traces that start from the same initial April 1 moisture conditions but apply different April-July weather sequences, drawn from 1981-2015 observations. (Source: CBRFC)

The CBRFC uses a conceptual (or simple dynamical) modeling framework for its seasonal forecasts, coupling a snow model (SNOW-17) with a rainfall-runoff model (Sacramento-Soil Moisture Accounting; Sac-SMA). Conceptual models draw from our understanding of the physics of the real-world watershed, but have very simplified representations of watershed attributes and processes, with 5-20 components and adjustable parameters to relate them to each other. CBRFC runs their models throughout the forecast season (December-July), using historical daily precipitation and temperatures from each year of 1981-2015 to stand in for the unknown future weather. This creates an ensemble of 35 forecast traces, each starting with the same observed initial moisture conditions but individually evolving the modeled snowpack and runoff through the end of the forecast period given the winter/spring weather of 1981, 1982...2014, 2015. These ESP (Ensemble Streamflow Prediction) forecasts are issued daily, and then become the primary input to the official forecasts issued monthly and semi-monthly.

NRCS uses a statistical modeling framework, employing separate regression models for each gage that are calibrated on the historical relationships between streamflow and several watershed predictors, typically snowpack (SWE) and accumulated water-year precipitation at 2-5 SNOTEL sites. Statistical modeling has been the standard-bearer for seasonal streamflow forecasting in the western U.S. for the past 50 years, and only very recently has its skill in the Colorado River Basin been exceeded by dynamical forecasts from CBRFC, which previously relied more heavily on statistical models similar to the NRCS framework.

Data and Tools

CBRFC

CBRFC has multiple ways to access and view their seasonal water supply forecasts and related data:

CBRFC Conditions Map - Water Supply Forecasts

CBRFC Conditions Map - Water Supply Forecasts

The Conditions Map can display several different types of hydroclimate information, depending on the tab selected. The Water Supply Forecasts tab, when opened, defaults to the most probable (50% exceedance) values for the official 1st of month forecasts, but can also display the latest daily ESP forecasts ("Latest Model Guidance"). Clicking on the triangles (forecast points) brings up a thumbnail forecast evolution plot for that point; clicking within the plot opens a page specific to that forecast point, with an interactive forecast evolution plot and links to plots for previous years, calibration and validation data, and historical streamflow volumes for that point.

CBRFC Upper Colorado Situational Awareness

This page shows the most recent official forecasts for Lake Powell inflows (~all Upper Basin runoff) for April-July and the water year; maps of observed/modeled soil moisture, snowpack, and precipitation used to inform the forecasts; and graphs showing how the sub-basins' contributions and overall forecasted volume compares with average conditions.

CBRFC Water Supply Official Forecast List

This page features an interactive table with the official 1st of month forecasts for all forecast points. Like the forecast map, it links to the pages for each point.

CBRFC Water Supply Forecast Discussion

This document (PDF) discusses provides a general overview of the official (1st and 15th of month) CBRFC forecasts, describing the weather/watershed conditions that have led to the forecasts, as well as forecasts and outlooks for upcoming weather.

CBRFC Water Supply Webinars

From January through May, CRBFC presents one-hour webinars around the 7th of the month to explain the latest streamflow forecasts, the weather and snow conditions leading up to the forecasts, and the weather conditions forecasted for the next two weeks. Scroll down to "Presentations 20xx" for a link to the webinar archive for the current year; and then find the Water Supply Webinar" of interest. Both the slides and a recorded video with slides are available.

NRCS

NRCS likewise has several ways to access and view their seasonal water supply forecasts and related data.

NRCS Interactive Map - Forecasted Streamflow Volumes

NRCS Interactive Map

The NRCS Interactive Map is a versatile tool to access and view all NRCS hydroclimate data, including the seasonal water supply forecasts. The link above has been specifically set up to show the most probable (50% exceedance) values for the current seasonal water supply forecasts, for individual points, but users can select other ways to visualize the forecast data, e.g., by basin, as percentiles, or for different forecast periods. Similar to the CBRFC Conditions Map, clicking on an individual forecast point will bring up a small window with the name of the point and forecasted streamflow volume; clicking on that window will bring up other options for viewing additional data for that point, including other statistics for the forecasted streamflow, and the observed streamflows for the water year.

NRCS Water Supply Forecast Chart

The Water Supply Forecast Chart shows the potential range for the forecasted streamflows (90%-70%-50%-30%-10% exceedances), similar to the shaded bands on the CBRFC forecast evolution plots. Note that the sub-basins within the CRB are shown by state; forecasts for Colorado R at Glen Canyon Dam (i.e., Lake Powell inflows) are accessed under Utah >Southwestern Utah.

NRCS State Basin Outlook Reports

The Basin Outlook Reports (also known as Water Supply Outlook Reports in some states) are issued monthly or semi-monthly from January through June by the state-level NRCS Snow Survey Offices, and supplement the NRCS seasonal water supply forecasts with narratives of statewide precipitation, snowpack, and water supply conditions, and additional graphics showing the water supply forecasts by sub-basin, alongside snowpack, precipitation, and reservoir conditions.

• Arizona (issued Jan-Apr)
• Colorado
• New Mexico
• Utah
• Wyoming

Additional Resources

State of the Science Report

Chapter 8 of the State of the Science report covers many aspects of Streamflow Forecasting in greater detail; Section 8.4 is specific to mid-range forecasts (seasonal and longer). Chapter 6, on Hydrologic Modeling, has a detailed description of the CBRFC forecast model framework in Section 6.3, under National Weather Service Models.

CBRFC Presentations

These two presentations from the October 2020 CBRFC Stakeholder Workshop provide more detail about CBRFC forecast procedures.

• CBRFC 101 - More general, covers both short-term and seasonal forecasts
• The Water Year in 45 Minutes - Specific to seasonal forecasts

Ongoing Research

There are three primary pathways to improving seasonal streamflow forecasts:

1. Improving the characterization and incorporation of the initial watershed moisture conditions (soil moisture and snowpack)
2. Narrowing the spread of potential trajectories of the future weather, e.g., by incorporating climate forecasts
3. Enhancing the structure of the forecast model itself, e.g., to expand the inputs or improve the model physics

CBRFC is involved in several recent and ongoing collaborative research projects in all three areas, including enhancing the snowpack data ingested into their model by using spatial SWE estimates, testing the use of seasonal climate forecast guidance (from NMME) to weight forecast traces, and testing the use of a more sophisticated snow model (iSNOBAL) in place of the current SNOW-17 model. At higher levels within NOAA, significant resources have gone into the development of a next-generation hydrologic forecast model (National Water Model; NWM), but it is not clear if or when the NWM will supplant the current model framework used for seasonal forecasting at CBRFC and the other NOAA RFCs.

The main research-to-operations effort for NRCS is their ongoing testing of a new statistical forecast framework based on an ensemble of machine-learning algorithms. This framework has more flexibility to identify useful relationships in the data, including non-linearities not handled well by standard regression modeling. The new framework has been run alongside the current operational approach to produce forecasts in several test basins for the 2020-21 season.