How to improve Objective Function

[MEN Vuthy]

Dear Sir, 

I want to simulate the model with data from 1998 to 2009. 

I used : 3 years for Warm Up Period.

            6 years (2001-2006) for Calibration

            3 years (2007-2009) for Validation

Please refer to attachment.

Could you please guide me how to increase the objective function of my calibration? Look at R_factor and P_factor are so conditioned, and also NSE is just 0.53.

I first did the sensitivity analysis by choosing 10 parameters and running one iteration with 500 simulations.

Then I just decided to choose only 9 parameters for further calibration with 2000 simulations at each iteration. I have tried to read to find out how to select parameters from sensitivity analysis, but I am still not so clear.

After 4 iterations of calibration, I got the results as shown in the attachments. And it seems the Obj Function doesn't seem to go better. 

I have tried validating using calibrated parameters from 4th iterations of calibration, and finally the Obj Funtion decreased to 0.43 of NSE, which is not so good.

I really have no more ideas on how to increase it. Please kindly help me.

Vuthy