Dear Alham,

do you have any measurements in the region?

Hydrologic models can only do so much.  I know of arid areas that lack summer rain, and they go dry, but I am sure there could be instances where they do not.  Most hydrological models do not consider groundwater flow, geology, springs, etc.  Here is a paper by TOTH that could help explain the potential complexity of groundwater movement, but in arid areas, another question may be is there enough to recharge.  But obviously, if the river is perennial, there is enough recharge or outside sources to supply it.  An interesting thing about groundwater is that it does not necessarily follow the surface hydrological boundaries.  I do not have any data to send relative to this model, but under the conditions you describe, zero flow seems appropriate. 

Mr.William thank you very much for this rationale. So we can say that the discharge output of ArcSWAT in this case is due to flood runoff that caused by rainfall. The river is perennial in its mean reach and there is enough recharge and some snow melt from a highland in the northern sides.

However, in the calibration step, I plan to calibrate the model using extrapolated observed daily total flow using before-development flow data (before 1950) to eliminate as possible factors other than climate can affect the river flow. So I wonder if it is may be better to calibrate using the flood flows instead of total flow? In which initially to separate the base flow from the flooding then calibrate! though I think using total flow seems appropriate!?. 

Dear Mr. Kleomenis Kalogeropoulos, which measurements are you referring to?