Kirchner response function

[Stefano Basso]

Hi all,

I started some time ago to look into shyft as a possible tool for a project.

After studying the documentation, one thing is not clear to me (I may have missed it):

- Does shyft set (and calibrate) a different Kirchner response function for each cell?

- Or does it use the same single response function for every cell?

- In the latter case, is it possible to set different response functions for different cells according to arbitrary criteria?

Thank you for the clarifications you may provide.

Best regards,

Stefano

[sigbjorn.helset]

Hi,
thanks for the interest, I will try my best to provide useful answers.

Q: Does shyft set (and calibrate) a different Kirchner response function for each cell?

A: Yes, it is possible, but the common usage is to identify/group cells together to form a catchment where you do have have observations that are common for those cells. E.g. observed discharge, snow SWE/coverage.
    The calibration process will then find the parameters that best matches the observation, given forcing data (precipitation, temperature, wind speed, radiation, relative  humidity).

    The calibration provides a flexible  set of tools to formulate the goal functions. (e.g. kling gupta,  nash sutchcliffe criiterias).

   -  For one specific stack, same set of response functions are used, but with parameters as obtained by calibration, as explained above. Kirchner also recommend using parameter estimation, which is outlined in his paper, thus a direct parameter estimation is possible.

   - Although it could be implemented, one specific  stack uses a specific set of response routines, (there are currently 5..7 different method stacks to select from).

Hope this helps.
Examples for calibration and run is provided in the examples/test routines in python, and even notebooks can be of help to get started.

Br Sigbjørn