ESTIMATION OF RUNOFF-ROUTING MODEL PARAMETERS USING INCOMPATIBLE STORM DATA

When several observed storm events are available a pooling method is required to estimate the parameters of a runoff-routing model. The usual pooling model makes the strong assumption that there is a set of model parameters unique to all storms. Using the RORB runoff-routing model in a case study, it is found that its model parameters inferred from one storm event are frequently statistically incompatible with parameters inferred from other storms. A more general pooling model is proposed, which views storm parameters as being random and independent samples from a random effect probability model; the random effects may be because of parameter bias arising from errors in catchment rainfall, or model simplification. The pooled mean and covariance matrix of the model parameters is derived in a Bayesian framework. The design hydrograph is computed using the pooled mean, which, unlike simple averaging, makes explicit allowance for storm data of varying quality. The computation of lower and upper bound prediction limits on the design hydrograph is discussed. It is found that the prediction limits can be sensitive to the choice of pooling model. © 1990.