Developing an ensemble hydrologic forecasting is difficult and time consuming work and choices are often required to determine how much effort is worth expending on one aspect versus another. Assuming an ensemble forecasting system uses ensemble weather forecasts, the key decision at hand is what other sources of uncertainty should be reflected in the ensemble forecast. Options include model structural uncertainty, model parameter uncertainty and model calibration period data uncertainty (eg, climate inputs or streamflow measurements). A related decision involves whether or not to utilize a distributed/semi-distributed hydrological model in the forecast system in place of an ensemble of easier to construct lumped parameter hydrologic models. On one hand, using an ensemble of lumped models enables consideration of model structural uncertainty in the forecasts. On the other hand, using a more …