Blow-up at first time step when ONLINE cppdef used

Hi there,

I am using BULK_FLUX and have already ran my model with ERA5 data interpolated offline. However, this is intensive on disk space so I would like to use the online interpolation option.

I have encountered an issue with define ONLINE where the model crashes after the first timestep. This is using the same data that was processed by the offline interpolation. When the model used the offline interpolated data, it ran for a full three years with no issue. Therefore, I didn’t expect it to crash now after one timestep.

Any comments or suggestions would be greatly appreciated.



Could you tell us how the model crashes ? what is the content of your croco.out file ?



Hi Matthieu,

I will come back to you, my model is now crashing in it’s original format (undef ONLINE) so there is something else wrong.


  1. There were only a few differences between the model setup with and without ONLINE activated.
    The main difference now compared to previously, is that I have adapted the contents of the file so that so that it contains the pressure field. This involved adding pressure to…
  1. …the create_bulk.m file, and processing pressure then as part of interp_ERA5.m.

However, when running ONLINE, these croco_blk files are needed for tides. I have added tides to the croco_blk files because there isn’t a croco_frc file when bulk formulation is used. This file is copied and…

  1. relabelled as as this is the file type that the tides must be read from. Is it possible that the model crashes, whether ONLINE is defined or not, due to the addition of the pressure field to the croco_blk files? If so, your advice would be appreciated on a suitable fix for this.

I think you are mixing the bulk file and the forcing file.
The tides waves must be contained in the forcing file (see forcing section in
The bulk flux are in the bulk files (see bulk_forcing in the

Thanks. I understand your point but I felt I needed to write the tide data to each croco_blk file (labelled by year and month). I have submitted a new question in relation to best practice for setting tides. In our operational work, tides are reinitialised annually. How is this done interannually?