Linear Models for Flow Estimation and Prediction 'Source Title' Modeling Hydrologic Processes, Proceedings Of 3rd International Hydrology Symposium
Bibliographic details
Summary
Linear models have been utilized for some years in hydrology to provide simple, easily applicable methods for modelling flow processes. This paper describes some of the main advantages and disadvantages of the linear approach to flow estimation and prediction problems. A typical problem is the need to estimate flow at various points throughout a branching river network if a point source of pollution is introduced, for example upstream of a river intake for public supply, and it becomes necessary to estimate its time of arrival at certain critical points in the river system. Another problem is examined in river regulation schemes where flow at a control point is maintained at some prespecified level then the prediction of flow some time ahead is required on an operational basis to facilitate the efficient management of the regulation reservoirs. The paper discusses in detail the applicability of multiple regression analysis, and an adaptive least squares algorithm, to the solution of the problems described whilst emphasizing the importance of devising efficient methods for the assessment of the performance of the various models produced. Detailed reference to recent British research will be made.