Simulation discharge different from calibration results

[Bikas C. Bhattarai]

Hello,

Using the exact parameters from the results of a calibration, I run a simulation over the same time period. When I extract the discharge using:

ts = simulator.region_model.statistics.discharge([])

I get different results than what is provided using the same following calibration. In both cases, the catchment ids are identical, and I am using the 'region_model.statistics.discharge' method to extract the data. 

Looking at the individual cell parameters confirms the parameters in both model configurations are identical. Any thoughts on what could be driving the differences and where to look for problems?

Following calibration:

Following simulation:

[Sven Decker]

Hei,

I get a very similar problem. I do exactly what Bikas is doing, but first my simulated discharge from the simulator is smoother compared to the discharge from calibration. My timesteps are equal. Also the peak discharges differ quite a bit:


Any ideas what is going on here?
Thanks,
Sven

[Yisak Abdella]

Hi

I get identical results when i carry out the same procedure you mention. Can u post the script u used.

Regards
Yisak

[Bikas C. Bhattarai]

Hello, 

I have attached my script for simulation and calibration herewith.

[John F. Burkhart]

After a long, but fruitful session, Felix and I have discovered that YamlCalibConfig does not seem to update the interpolation parameters given in the yaml files as the YamlSimConfig does.

This was leading to a nearly factor of two difference between the interpolated precipitation resulting from IDW parameter differences.

We will make an issue to address this, but it would probably be more fruitful to move ahead with externalizing the SK orchestration routines as discussed.