The deep ocean layers 2-9 start at the annual maximal mixed layer depth. The thickness of layer 2 (DZ2) is 10m and the layer thicknesses increase geometrically s.t.
DZ1X+DZ2+DZ3+...+DZlmom=1000m (lmom=9)
i.e.
DZ1X=10/Q DZ3=10*Q DZ4=DZ3*Q, ... and DZ1=ann.max.mixed layer depth
DZ1X is not used; the dzo's to compute the vertical gradients are given by:
dzo1=10/sqrt(Q) dzo2=dzo1*Q dzo3=dzo2*Q etc. Q=1.701...
The diffusion into the deep ocean is computed once every day using the temperature anomaly with respect to 'climatology' as tracer. The climatology is found by taking a 10-year mean of the Qflux model without a deep ocean that is used to initialize the deep ocean run.
More detailed description:
The heat flux at interface of layers L and L+1 is given by
ED*(RTGO(L)-RTGO(L+1))/dzo(L) L=1,...,8
however, a semi-implicit scheme is used to compute these fluxes.
The following auxiliary arrays are used:
Note: After 1 year we have
ADTG3 = ( SumD T3_yr1 - SumM init.TG3M )/365
After 2 years:
SumM TG3M = SumD T3_yr1
hence
DTG3 = SumD T3_yr1 - SumM init.TG3M + SumD(T3_yr2 - T3_yr1) ADTG3 = ( SumD T3_yr2 - SumM init.TG3M )/365
i.e. the year 1 contributions canceled
Compared to the Qflux model without deep ocean, the model with deep ocean needs one extra input file (ED.. diffusivity coefficients).
The code needs one call to ODIFS (at the end of each day). ODIFS calls ODFFUS (older name DIFFUS). The common block OCN containing TG3M,RTGO,STG3,DTG3 is needed in main and input, since these arrays have to be added to the restart file, RTGO(.,.,2-lmom) was also added to the acc-file for convenience. RTGO,STG3,DTG3 have to be set to 0 initially in INPUT, TG3M has to be computed from a preliminary run (use KCOPY=3 in the preliminary run) and added to the IC-file.