Testing and Improving Distributed Runoff Models for Flood Forecasting as part of the National Weather Service (NWS) Distributed Hydrologic Modeling Intercomparison Project Phase 2 (DMIP 2)
The basic problem: short-term (one to five days), real-time flood runoff forecasts are generally inaccurate due to precipitation and runoff estimation issues. Improved flood runoff forecasts help water-supply decisionmakers best allocate scarce resources and improve joint use and flood control storage at existing Reclamation reservoirs.
The main research question is:
* Can distributed hydrologic models provide increased simulation and forecast accuracy compared to lumped models?
* Also, what is the performance of distributed models if they are calibrated with observed precipitation data but use precipitation forecasts?
These questions will be investigated by Reclamation as part of the DMIP 2, led by the Hydrology Laboratory (HL) of the National Weather Service (NWS). Reclamation will test and improve the TREX distributed model on the North Fork American and East Fork Carson Rivers for flood forecasting under DMIP 2.
Need and Benefit
This product is a technical tool that can be used by engineers and scientists to improve short-term operational forecasts. Reclamation has a need for this research product in order to potentially improve short-term flood runoff forecast decisions during flood operations at Reclamation reservoirs. This research product is a tested and improved distributed rainfall-runoff model that will hopefully provide better flood runoff forecasts. The tool is designed to be deployed in conjunction with NWS River Forecast Centers (RFCs) in the Western United States.
This research does have potential for broad application across Reclamation. Since Reclamation currently depends on NWS flood runoff forecasts, it is useful and prudent for Reclamation to participate and help implement an improved tool so that it works for Reclamation and meets our needs, in addition to those of the NWS. The research approach that will be used is to start with a specific location and generalize the tools for application Reclamation-wide. The initial sites for testing the tools are within the Mid-Pacific (MP) Region. Initial testing is proposed on the North Fork American River in the American River Basin (Central Valley Project-Folsom Dam) and the East Fork Carson River in the Carson River Basin (Lahontan Dam). The tool could eventually be applied at other Reclamation dams in conjunction with NWS RFCs, especially the California-Nevada and Colorado Basin RFCs.
The distributed modeling tool will potentially help contribute to increasing water availability and cost avoidance by providing decisionmakers improved quantitative short-term flood forecasts for reservoir operations. The flood runoff predictions can help decisionmakers in two scenarios: The first is to better coordinate releases from storage to accommodate large inflows and avoid downstream infrastructure damage, thereby avoiding potential costs. The second scenario is to determine that the reservoir has sufficient storage to contain the flood runoff volume and store this water for later use in the irrigation season. In these ways, Reclamation can potentially save thousands of dollars in better managing flood operations releases. The overall goal is to partner with the NWS so that Reclamation benefits from improved forecasts.
There are existing capabilities available to Reclamation from external sources that answer these short-term flood forecasting questions at some Reclamation facilities. The existing capabilities near the proposed research site are part of the NWS California-Nevada RFC. The NWS closely and constantly coordinates forecasts and shares data with Reclamation in Sacramento, California. The NWS is actively conducting research as part of the DMIP to improve their forecast models, and looks to eventually replace or improve the existing Sacramento model. This research proposal seeks to enhance the relationships and coordination between Reclamation, the NWS HL and forecasters by participating in their modeling experiments on a watershed relevant to Reclamation and sharing Reclamations' expertise in this class of models.
If the proposed tools were unavailable, existing NWS rainfall-runoff models would be used in their current capacity. The consequences would be less accurate, three-day reservoir inflow flood hydrograph estimates. This end product will be an improvement over existing capabilities in improved runoff modeling using distributed, physically-based methods.
Contact the Principal Investigator for information about these documents.