Adjustments to Short-Term Hydrologic Forecasts for Conceptual Models of Hydrologic Processes

by Norman Crawford,

Document Type: Proceeding Paper

Part of: WRPMD'99: Preparing for the 21st Century


(No paper) Real-time forecasting of streamflows is amorphous and multi-dimensional. Weather forecasts, which are synonymous in popular culture with uncertainty, are just one component of a streamflow forecast. A flood forecasting system includes; 1) field data stations that report weather, streamflows, and reservoir levels in real-time up to the present time or NOW. 2) current weather forecasts, especially for precipitation and temperature 3) physically based algorithms for modeling hydrologic processes 4) data base systems to handle real-time and forecast hydrometeorologic time series, and hydrologic model state variables (snowpacks, soil moistures, reservoir contents). Forecasts are made quickly. Available data are often incomplete or unrepresentative. Data that will be obviously erroneous in hindsight often seems legitimate in real-time. Differences between model streamflow and actual streamflow measured at NOW needs to be used to improve forecasts. When physically based algorithms are used in a conceptual model the state variables in the model may be updated. Updates to state variables must be reasonable so that the advantages of physically based algorithms are maintained. Incorrect updates caused by noisy real-time data must be minimized. A strategy for making state variable updates is discussed, and the effects of this strategy are illustrated for rivers in Maine and Connecticut.

Subject Headings: Hydrologic models | Forecasting | Streamflow | Algorithms | Floods | Hydrologic data | Reservoirs | Water demand | United States | Maine | Connecticut

Services: Buy this book/Buy this article


Return to search