Water Supply Forecast Rodeo: Forecast Stage

[submissions closed] Water managers in the Western U.S. rely on accurate water supply forecasts to better operate facilities and mitigate drought. Help the Bureau of Reclamation improve seasonal water supply estimates in this probabilistic forecasting challenge! [Forecast Stage] #climate

$50,000 in prizes
9 weeks left
47 joined

Project background

Accurate seasonal water supply forecasts are integral to water resources management in the Western United States, where conditions are dry and there is high demand for water. Water managers use supply forecasts to inform a broad range of decisions related to water supply management, flood control, hydropower, and environmental objectives.

Photograph of Hungry Horse Dam in Montana.
Photograph of Hungry Horse Dam in Montana. Image courtesy of USBR.

Since the 1920s, forecasts of water supply have been used to guide and inform water management. Early methods of forecasting water supply involved estimating the water volume in manually measured mountain snowpack. Modern prediction systems still rely in part on mountain snowpack measurements and related estimates, like river runoff volumes, but also incorporate information about other hydrologic processes like evapotranspiration from crops and forests, and ground infiltration of rainfall and snowmelt. The newest forecast products leverage recent advances in machine learning, AI, and applied statistics, and incorporate broad sets of features from varied sources like observational datasets.

In this challenge, solvers will explore ways to improve seasonal water supply forecasts, advance forecast explainability, and characterize uncertainty of forecasts. Avenues for innovation could include the application of novel machine learning methods or approaches in modeling that account for the complex, dynamic nature of hydrologic processes. For example, solvers may incorporate new data sources, such as observational data, climate forecasts, or estimates of basin conditions, and may find or engineer features related to seasonal runoff and changing land cover conditions due to events like prolonged drought/heat stress, pests (e.g., mountain pine beetle), wildfire, land conversion, development, restoration, and land impacts resulting from climate change.

Improvements to water supply forecasts greatly benefit water managers, who are working with increasingly narrow margins. Climate change and a reduction of snowpack are decreasing water supply, just as demand for water is increasing due to population growth. Getting the best seasonal forecast information possible will better equip water resources managers to operate facilities for high flows, mitigate impacts of drought, improve hydropower generation, and meet environmental targets.

Existing models

When modeling seasonal volume, the National Weather Service River Forecast Centers (RFCs) try to model "natural" flow, or flow that would be expected given no water management activities. One popular technique to model cumulative natural flow for various regions is called Ensemble Streamflow Prediction (ESP). ESP is a part of the National Weather Service Community Hydrologic Forecast System and produces probabilistic forecasts of hydrological variables. It simulates soil moisture, snowpack, regulation, and streamflow, using current conditions and historical meteorological data to generate likely future hydrological scenarios. These forecasts aim to predict adjusted water volumes, considering upstream storage changes and consumptive use.

Another method called Statistical Water Supply (SWS) is a software program that uses regression equations that relate observed data to future seasonal streamflow volume. It utilizes monthly inputs, including total precipitation (across multiple months), snow water equivalent, monthly flow volume, and climate indices like the Southern Oscillation Index (SOI). The output of this method is a conditional probability distribution for a seasonal volume, typically spanning from April to July. Instead of a single value, the result of the equation represents the 50% exceedance level. Different exceedance levels, such as 10% or 90%, can be calculated using the standard error, offering a probabilistic perspective on future streamflow. The Natural Resources Conservation Service (NRCS), a part of the U.S. Department of Agriculture, issues statistical water supply forecasts for hundreds of locations in the Western U.S.

About the Bureau of Reclamation

The Bureau of Reclamation (Reclamation) is an agency within the United States Department of the Interior that has worked since 1902 to manage and deliver reliable water and hydropower for the Western United States. In service of their mission, Reclamation builds and maintains infrastructure like dams, reservoirs, power plants, and canals, including including Hoover Dam on the Colorado River and Grand Coulee on the Columbia River.

Reclamation operates and maintains water and power projects in 17 Western U.S. states:

Map showing regions where the Bureau of Reclamation operates.
Map of the Bureau of Reclamation regions. Image courtesy of USBR.

Reclamation sets environmental targets as part of water management, and also works to protect and restore aquatic ecosystems and wildlife habitats. In recent years, water conservation and efficiency are a priority for Reclamation. Efforts to reduce water waste and increase the use of recycled water align with broader sustainability initiatives for the Bureau.

Reclamation is the largest wholesaler of water in the country. It brings water to more than 31 million people, and provides one out of five Western farmers (140,000) with irrigation water for 10 million acres of farmland that produce 60% of the nation's vegetables and 25% of its fruits and nuts. Reclamation is also the second largest producer of hydroelectric power in the US. Its 53 power plants annually provide more than 40 billion kilowatt hours generating nearly a billion dollars in power revenues, and representing 15% of the nation's hydropower.

You can learn more here

Development Arena thumbnail image is a photo of Yellowstone River near Yellowstone National Park. Courtesy of USBR.