Reed, D.W. 1984, reprinted 1986. A review of British flood forecasting practice. Report No. 90, Institute of Hydrology, Wallingford, 113 pp.
To read the report's recommendations: Click here
The feasibility and success of a flood warning scheme depend on many factors, of which data acquisition and procedures for disseminating warnings are particularly important. Although the review examines these topics in some detail, the core of the material presented refers to the application of flood routing and rainfall/runoff models to flood forecasting. The continued use is recorded of simple correlation techniques for flood routing. The report describes the Muskingum family of flood routing methods and explains the particular merits of the variable parameter Muskingum-Cunge technique. Rainfall/ runoff methods are reviewed under four headings: unit hydrograph techniques, nonlinear storage methods, transfer function methods, and conceptual models. Nonlinear storage models are seen to be well suited for flood forecasting; they are relatively simple in structure, not too difficult to calibrate, and can be readily implemented. The transfer function approach is explained in simple language and its particular attributes for real time forecasting compared to those of the more familiar unit hydrograph approach. The review considers in detail the problem of how best to correct flood forecasts by reference to telemetered flow measurements. Three methods are distinguished: error prediction, state-updating, and parameter-updating. Other topics reviewed include the role of weather radar, the particular obstacles faced in flood warning on rapidly responding catchments, and general guidance in matching a forecasting technique to a particular catchment. The report cites more than 120 references.