Storm surge modelling - extremely high wind drag coefficients? - Forum - Delft3D Flexible Mesh

Dear all,
I run a model of the Venetian lagoon, which has been calibrated against for the tide based on a extended period without the influence of wind (see tidalcalibration.png). Performance is really good. However, I want to actually run the model for very stormy conditions. When taking into account a uniform wind- and pressure field over the model domain which had been done in other studies as well the default linearly varying drag coefficient (cf. Smith and Banke (1975)) underestimate the peak by about 20 cms (see StormEvent_windDrag_adjustment.png). I have to change the wind drag coefficient for 100 m/s from 0.00723 to 0.0123 in order to be able to reproduce the peak. 

However, when running a validation calculation for another storm event (see tidal_signal_PuntaSallute.png) the chosen wind drag coefficients seem to overestimate the peaks.

I am now trying to find possible reasons why I need so extremely high wind drag coefficients in the first place. Does anybody know possible reasons for this?
I checked my boundary forcing (identical with the ones used in other studies), bathymetry (make sure to use correct reference datum), and bottom friction. I define the wind input by means of a uniform wind speed and direction .tim file. Same for the pressure.

Looking forward to your thoughts on this,
Thank you and all the best,
Julius