Up to Specific issues in real coupled models
Hi, I wondered if anyone has any advice on or experience of coupling a single scalar value between two components using OASIS3-MCT (or OASIS3). A colleague would like to exchange as single, non-gridded real value between two model components at the same frequency as the rest of the "normal" coupling fields using OASIS3-MCT. I'm not sure whether it's really a good idea to go down this route, and I'm not quite sure how it would be achieved (maybe define an identical single point "grid" in each component to accommodate the value in question?) but if anyone has any thoughts about : a) how to exchange a single value b) whether it's a good or bad idea try to exchange a single value using OASIS3(MCT) I would be grateful to hear your thoughts or experiences. Thanks, Richard
Hi Richard, You could try to define two identical grids with one point in each model (see the points partition in the User Guide https://oasis.cerfacs.fr/wp-content/uploads/sites/114/2021/02/GLOBC-TR-oasis3mct_UserGuide3.0_052015.pdf, section 2.2.3) and exchange your value through this point as it is not possible with OASIS to exchange data if they are not defined on a grid. Let me know if this work. Best regards, Laure
Hello Richard and Laure, What are the conclusions about this topic ? Richard, did you test to define one point grid to exchange scalar data ? In this case, one specific source process exchanges the scalar value to one specific target process, right ? Is there any way to exchange a scalar value from all of source processes to all of target processes ? From one specific source process to all of target processes ? Thanks in advance ! Arnaud
Hi Arnaud, I have not personally tried the single grid point approach myself, or from one source grid point to all target points. However I would like to try either of those because I think they would be a better solution than the one which has been implemented here. That approach (to pass a single ice sheet mass balance value from our atmosphere to NEMO-CICE) populates an entire global array with the same identical value, then passes it via the coupler to NEMO. The problem with that is that obviously NEMO receives thousands of slightly different variations of the same number due to rounding discrepancies, so once NEMO has received them, it does a bit-reproducible global sum and then takes the average value by dividing the sum by the number of grid boxes! It does seem to work adequately but it's messy and I would like to replace it with one of the approaches you mention. Richard
