Hello,
OK. The only improvement I see would be :
TRAITEMENT DES HAUTEURS NEGATIVES : 2
to have strictly positive depths, this could help k-epsilon.
Also you do not need to have :
DISCRETISATIONS EN ESPACE : 12 ; 11 just try 11 ; 11 instead.
You can add the diffusion coefficient in the variables for graphic printout, so that we see what kind of diffusion gives the k-epsilon model. You can also try to change EMAX and KMAX (maximum epsilon and k) in subroutine cosake.f, trying 100. instead of 1.D10.
If it is not already done, make sure that your model works without algae and with a constant diffusion, let's say = 1 m2/s.
With best regards,
Jean-Michel Hervouet