Python version issues

[jarethholt]

Hi Fabien,

I noticed two things in the Python version. First, the 75-term EOS uses a salinity offset of 32 g/kg instead of 24, so its results don't match the reference values. Second, there are some slight inconsistencies between the expansion and contraction coefficients calculated with the given polynomial coefficients versus directly calculating the density/volume polynomial derivatives. This could just be floating point errors made when these coefficients were copied, but it can affect the physical quantities by a few percent or more.

I've been interested in Python versions of TEOS10 for a while, and posted my take on polyTEOS10 here. Let me know what you think; I would be more than happy to merge it here.

[fabien-roquet]

Hello Jareth, You are right about the python version for the 75t where deltaS should be 24 instead of 32. Thank you very much for pointing out this mistake. I have made the correction on GitHub. As for the alpha/beta coefficients, they should be correct however it’s been awhile I have not checked. The matlab code has been more thoroughly tested so I would trust it 100% while the python code could have some bug left unfortunately. There are different version of TEOS-10 and my 75-term version has been proposed as part of TEOS-10 for a while now. There is in particular a python version (using a C wrapper): https://github.com/TEOS-10 https://github.com/TEOS-10/GSW-Python I have not checked it but I would assume that the work has been done correctly. Maybe you’ll find this version useful, as it is including a lot more than the equation of state. You can maybe compare your version with the TEOS-10 one to see if it’s all consistent. Thanks a lot! Fabien

…

On May 15, 2019, at 5:47 PM, Jareth Holt ***@***.***> wrote: Hi Fabien, I noticed two things in the Python version. First, the 75-term EOS uses a salinity offset of 32 g/kg instead of 24, so its results don't match the reference values. Second, there are some slight inconsistencies between the expansion and contraction coefficients calculated with the given polynomial coefficients versus directly calculating the density/volume polynomial derivatives. This could just be floating point errors made when these coefficients were copied, but it can affect the physical quantities by a few percent or more. I've been interested in Python versions of TEOS10 for a while, and posted my take on polyTEOS10 here <https://github.com/jarethholt/polyTEOS10>. Let me know what you think; I would be more than happy to merge it here. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#1?email_source=notifications&email_token=AB653ERD5CN2J27D3D2NJMLPVQWCHA5CNFSM4HNEUUI2YY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4GT6X7KQ>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AB653ETLL4B3DXLFL4XLAOTPVQWCHANCNFSM4HNEUUIQ>.