Estimating uncertainty in hydrologic model predictions

Parameters for physically based hydrologic models are typically developed based upon user judgement and look up tables. Parameter estimates inherently contain a certain amount of error that affect the reliability of hydrologic models. By considering these estimates to be random variables, the output of the models in which they are used must also be considered as uncertain. While sensitivity analysis provides information about how much the output changes with small changes in a particular input parameter, uncertainty analysis goes further. It provides the statistical properties and an estimate of the statistical distribution of the output from the statistics or distribution of the input. Thus, the error introduced into model results by uncertain model input parameters can be assessed quantitatively.