Application of the precipitation-runoff modeling system to small basins in the Parachute Creek basin, Colorado

The U.S. Geological Survey 's Precipitation-Runoff Modeling System was calibrated and verified for two small basins in western Colorado. Average monthly mean, average monthly maximum, average annual total, and average daily mean observed and model predicted streamflow differences were small, on the average < 10 %, indicating the Precipitation-Runoff Modeling System can, when adequately calibrated, satisfactorily estimate these streamflow factors. Average monthly minimum streamflow generally was over-predicted, but actual differences between observed and model predicted average monthly minimum streamflow are small. To determine relative errors of streamflow prediction in ungaged basins, three different model parameter sets were used to predict streamflow in a nearby gaged basin. The first set of parameters was based on the assumption that little information about this basin was available. Model parameters used to calibrate basins approximately 75 mi away were adjusted to model this basin on the basis of easily obtainable physical basin characteristics, such as elevation and slope. The second set of parameters was based on the first set, except it was assumed that local climate information was available to estimate the model 's climatic parameters for the basin better. The second set of parameters improved streamflow estimation in this basin by about 7% over the first model parameter set. The third set of model parameters assumed that the model had been calibrated to similar, nearby basins. This model parameter set was derived from the model parameters of the earlier calibrated basins. Using these calibrated model parameter values improved streamflow estimates by approximately 21% over the second parameter set. The average difference between observed and predicted average annual streamflow for the basin using these parameters was 16.3%. For comparison, the average difference between observed and predicted average annual total streamflow for the calibration basins was 12.9%. This suggests that, with some increase in error, this model can be used to estimate streamflow in ungaged basins, if the model has been calibrated to a nearby gaged basin with similar physical characteristics. (Lantz-PTT)