Model Calibration and Sensitivity analysis

[Saeed Golian]

Hi everyone,

I am trying to do sensitivity analysis and model calibration using the attached params, but for both phases it give me error messages, sometimes the model totally crashes and the window disappears and I am not sure what the reason might be.

I am using a gridded version of the model. 

Regards

Saeed

[Saeed Golian]

A surprising point is that on the first try in Sensitivity analysis, the sampled CN value is negative! Why is that happening? I set cn2 to change by Percent with -30 to 40 percent range.

[Celray James]

I think I can help check what is going wrong. Do you mind sending  an sptp of your project?

Export project using the export section, upload to https://swatplus.link/share

send the link so I can check what is happening. I am sure there is a parameter that is causing issues.

Thanks

[Saeed Golian]

Hi Celray,

I just uploaded the project file. I’d really appreciate your help. 

Regarding the sensitivity analysis or auto-calibration, isn’t there any possibility to add some constraints, so if sampled values are beyond the feasible range, then the upper/lower limit be substituted (depends on which side the sampled value is)?

Kind regards

Saeed

--
You received this message because you are subscribed to the Google Groups "SWAT+ Toolbox" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sptoolbox+...@googlegroups.com.
To view this discussion, visit https://groups.google.com/d/msgid/sptoolbox/722350c3-03b8-4bac-8918-8fb12a156b0dn%40googlegroups.com.

[Celray James]

I got your project. I am traveling right now and my laptop will have serious trouble handling a run with that many HRUs. But I will give it a try. Have you considered trying a lower resolution (~ 1500 HRUs Max) before going all in on high resolution?

Concerning out of bounds parameters, this is supposed to be corrected by cal_parms.cal within the SWAT+ source code executable. Outside of that, it would be very complicated especially for spatially varied parameters such as CN2 using relative change and percent change.

[Saeed Golian]

Hi Celray,

When I try to calibrate the model, nothing's happening at all, after several hours, I cannot even see the 'Dotty plots' which shows at least initial sampling has been done! Can I ask if the calibration period must be the same as simulation period? For example the calibration data period is 01/08/2014 to 03/10/2024 but simulations (including warm-up period) covers the period 01/01/2023 to 15/10/2024. 

I also found the simulated GW table data are far from observed boreholes' values. I found the initial water table depth (dep_wt) should range between 0-10m, in my case they ranges from 3-4m to 20-21m. The same for aquifer depth (depth to bedrock, dep_bot), the model suggests the value to range between 0-10m while my actual values are around 30m. 

I changed the values of dep_wt and dep_bot to actual ones in  aquifer.con and aquifer.aqu files. Do you thing the reason the SWAT+Toolbox is unable to even start calibration relates to these changes (which I don't think so as I can run the model using these new values for gw depth and aquifer bottom in both SWAT+Toolbox and Editor). I already sent you my project, attached please find the time series of observed GW level (for aqu25442).

Thanks in advance

Regards

Saeed