The application of the model in the Filefjell Basin led us to some conclution concering tho model structure and the relative importance of its parameters.
The original HBV-2 model had to be modified with the introduction of a third runoff component. A oompletely lumped modifioation proved to be better than a more distributed one.
Once the model was modiified the calibration was relatively simple.The most important parameters were those in the snow-routine and the recession coefficiants in the responee function. One of the biggest problems when trying to apply the model to ungauged catohments seems to bo the empirical coefficiente that we are forced to use when extrapolating preoipitation observations to areal values.
The soil moisture routine is less important in this alpine catohment because evaporation is low. A fictive value of the available water, Fc, of 150 mm was assumed but the maximum deficit in the soil moisture zone in the model was less than 50 mm during the seven years in the period.
The conflict between subjective visual inspection and optimum parameters in the least squares-sense in a strong indication that fittingsoriterious must not be used blindly, unless their capability of representing tbe agreement between tbe observed and computed hydrographs is explored rlgorously.