Strange convergence curve: seemingly converged but objective function still fluctuates

[rogerzzl]

Hi there,

I am using spotpy version 1.6.2 for parameter calibration of the VIC model with the SCE-UA optimization algorithm. My objective function is set as 1-NSE, which I aim to minimize. I set the maximum number of runs to 5000, but the process always stops after a little more than 3000 iterations. I assume this is due to early stopping after convergence.

However, when I visualize the objective function, I notice that even near what appears to be the convergence baseline, there are still many new parameter trials causing the objective function to jump and fluctuate, rather than flattening out into a typical horizontal convergence line.

I would greatly appreciate any advice or suggestions on how to address or interpret this behavior. Thank you in advance for your help!

[krishnakafle]

I am also facing the same issue. I am using KGE as an optimizer.

    def objectivefunction(self, simulation, evaluation):

        # Calculate Kling-Gupta Efficiency
        kge_value = spotpy.objectivefunctions.kge(evaluation, simulation)
        return kge_value

If you found the solution, please let me know.