Modification in COARE 3.5 source code

[Shun Ohishi]

Dear Dr. L. Brodeau,

I found that ztmp0=0 might induce zeta_u=NaN in the following code in mod_blk_coare.f90:

  zeta_u = (1.-ztmp1) * (ztmp0/(1.+ztmp2/(-zu/(zi0*0.004*Beta0**3)))) &
     &  +     ztmp1   * (ztmp0*(1. + 27./9.*ztmp2/ztmp0))

I suggest to replace from the latter part

ztmp1 * (ztmp0*(1. + 27./9.*ztmp2/ztmp0))

to

ztmp1 * (ztmp0 + 27./9.*ztmp2)

This problem would be dependent on fortran compiler,
because it occurs after changing my compiler.

Best regards,

Shun Ohishi

[Laurent Brodeau]

Hello,

First, mod_blk_coare.f90 does not exist any longer... You seem to be using a very old version of the code and I strongly encourage you to update to the current version! >> git pull !

But otherwise you are completely right, it was bad coding, and this actually even helped me to spot and fix a bug. I just committed the changes, it affects these modules: mod_blk_ecmwf.f90, mod_blk_coare3p0.f90, mod_blk_coare3p6.f90.

Cheers, /laurent

[Shun Ohishi]

Hi,

I found another bug in the former part in the same sentence, but the cause is not clear.

 zeta_u = (1.-ztmp1) * (ztmp0/(1.+ztmp2/(-zu/(zi0*0.004*Beta0**3)))) &

should be replaced as

 zeta_u = (1.-ztmp1) * (ztmp0*zu/(zu-ztmp2*zi0*0.004*Beta0**3)) &

Thanks,

Shun Ohishi