Issue with the interpolation between time steps in forcing file

[vdu001]

Hello,

I have encountered some issues related to how ROMS interpolates between input timesteps from the forcing file.

I am running model fields where I only want to change the surface stress with time in the forcing file. SST, SSS, and dQdSST had the same fields at the start and the end of the model run. To set this up, I let the surface stress (sustr/svstr) have several timesteps to allow these variables to change with time, while for SST, SSS, and dQdSST I only set up two time steps: one at the beginning and one long after I knew the run should be done. As far as I understand ROMS should then interpolate between these two timesteps according to the nudging assigned. Our run is 150 days long, and our nudging time scale is 18 days.

However, when I run the model with this setup it turns out differently from when I run the model with an additional timestep although the field in this time step is exactly the same as the fields at the start and the beginning.

As a simplified example: if I run the model with for instance wind at t_start=3m/s and t_end=3m/s, this does not give the same result as if I also assign wind=3 m/s at a time step between t_start and t_end.

I'm surprised by this result. Has anyone else encountered a similar situation?

All the best, Vår Dundas and Kjersti Daae

[jbettencourt]

If you want to use constant forcing fields, you can define only one time record for the fields you want to keep constant during the simulation.

[vdu001]

Hello, thank you for your response!

However, I do want to use varying forcing fields. The example with the constant forcing was just to explain the issue: that the result of the simulation is dependent on the number of time steps of the forcing file even when the forcing is the same at all time steps.
This result does not agree with my understanding that the model interpolates between forcing input according to the nudging assigned. This might affect our simulations, and I'm wondering whether the model could somehow interpolate or read the forcing file differently than expected?

All the best, Vår

[jcwarner]

seems odd.
when you did the test with the extra time step in the middle, what did roms report out when it read the file?
it always reports out the min, max, times, etc. check that.