Peak Discharge Difference between single lumped basin model and distributed model

[kiwi]

Hi, All
I am currently modelling a 46km2 urban/rural catchment using HEC-HMS using the SCS unit hydrograph method. What we have noticed is the peak discharge from the model can vastly different between a lumped single basin model and a distributed model representing the catchment using about 90 subbasins. The distributed model give a peak discharge about 90%^higher than the single basin model. I know there is something to do with the lag time, but in both model we have used the accepted method taking into account the true slope and the longest flow path etc.

Anybody with similar experience can shed a light on this?