Request for Guidance on Sensitivity Analysis, Calibration, and Validation in SWAT+ Toolbox (v3.1.0)
Chandu Palakuri
Dear SWAT+ Toolbox Team,
Good evening.
My name is Chandu, and I am currently conducting a Water Availability Assessment using the SWAT+ Model. I am in the calibration and validation phase and using the SWAT+ Toolbox (version 3.1.0) for this purpose.
For the sensitivity analysis, I have selected 18 parameters covering a calibration period from 1997 to 2015, with a 3-year warm-up period. I am using the Sobol method with a seed value of 100, 3800 samples, and 8 parallel processes. However, the model run takes nearly a full day to complete.
I would like to ask for your guidance on the following points:
-
Optimal number of samples to obtain reliable and stable sensitivity results while maintaining reasonable computational time for the Sobol method.
-
I am using the Dynamically Dimensioned Search (DDS) algorithm for calibration with a maximum of 1000 iterations. Could you please suggest an appropriate number of iterations that would help achieve good calibration performance without excessively long run times?
-
If possible, could you please explain the recommended workflow for sensitivity analysis, calibration, and validation in the SWAT+ Toolbox?
-
Additionally, if there is any official manual or detailed documentation available for the latest version (v3.1.0), I would greatly appreciate it if you could share the link or resource.
I have noticed that even after running the model for nearly a week with the above settings, the results are not satisfactory. Hence, I would like to confirm if my setup parameters are appropriate.
Thank you very much for your time and support. Any suggestions or resources you could provide would be very helpful for improving my workflow and model performance.
Kind regards,
Chandu Palakuri
M.Tech – Land and Water Resources Engineering
IIT KHARAGPUR