IMPACT OF RANDOM DATA ERRORS ON PARAMETER-ESTIMATION IN HYDROLOGIC MODELING

Parameter uncertainty in hydrologic models is due, in part, to random errors in input data used for calibration. This work investigates the impact of various error distributions associated with input data on the final estimated parameter values using three different estimation criteria. Errors in precipitation data were found to introduce more uncertainty into parameter estimates than errors in runoff data. Parameter uncertainty increased as the level of error introduced into input data increased. Correlated errors in the input data greatly increased the uncertainty associated with parameter estimates.