Modelling Future Streamflow

[Erika]

Dear SWAT+ Group,

I would like to ask for your advice regarding a modelling workflow I am currently working on.

I set up my model in SWAT+ Editor and performed the calibration using SWATrunR. From the calibration, I identified a set of parameters corresponding to the best-performing runs. During the model setup, I used the Hargreaves method for evapotranspiration estimation. And streamflow is my main variable of interest. The calibration and validation were based on observed data up to 2024.

My current objective is to use climate projection data together with the calibrated parameter set to simulate future streamflow. To do this, I created copies of my baseline model and replaced the observed climate inputs with projected climate data, creating separate model versions for each ensemble. I then attempted to run these models in SWATrunR using the calibrated parameter set to obtain future streamflow projections for the catchment.

The model run, however, the resulting streamflow does not appear reasonable, which makes me question whether my conceptual approach is correct.

Could you please share your opinion on this workflow? I'm interested to know:

• Is it appropriate to use copies of the calibrated baseline model with projected climate inputs in this way?

• Are there any common pitfalls when applying calibrated parameters to future climate scenarios in SWAT+ / SWATrunR that I should be aware of?

If you have experience with a similar workflow, I would greatly appreciate any guidance or suggestions.

Thank you very much for your time and support.

Have a great day :)

Erika

[Natalja C.]

Hello Erika,

You followed a workflow, and I do not see any major issues, but the:

-  "projected climate data, creating separate model versions for each  ensemble" --> you need to create new wgens (weather generators) for your projections as well. Do not forget to also update the CO2 for your future projections. I suggest using tools: https://github.com/biopsichas/SWATtunR and https://biopsichas.github.io/SWATprepR/index.html . Do not forget to project all your inputs, such as crop/land management (rotations), point sources, etc. etc.

- A good practice to ensure your projected weather data is "good", you can simply expend your current weather data with new entries. If the model produces expected results for your observation data and unexpended during the projections - that means that your weather projections are not correct. Consider re-doing downscaling  and/or bias correction.

To answer your questions:

• Is it appropriate to use copies of the calibrated baseline model with projected climate inputs in this way? --> sure! This is the way I do it, but include all the projected datasets (point sources, different wgens, co2 change, atmospheric deposition, extended management)

• Are there any common pitfalls when applying calibrated parameters to future climate scenarios in SWAT+ / SWATrunR that I should be aware of? --> the one mentioned above.

Good luck!

Natalja