Package longstep

This package allows the passive tracer time step to be longer than that for dynamical fields: the ptracers are updated only every LS_nIter time step. Dynamical fields are averaged over LS_nIter time steps and are available as fields LS_* (declared in LONGSTEP.h):

original fld. averaged fld.

UVEL LS_uVel VVEL LS_vVel WVEL LS_wVel THETA LS_theta SALT LS_salt IVDConvCount LS_IVDConvCount Qsw LS_Qsw

Kwx LS_Kwx Kwy LS_Kwy Kwz LS_Kwz

KPPdiffKzS LS_KPPdiffKzS KPPghat LS_KPPghat

The T and S time step remains the same as that for u,v,...

Packages that use ptracers (like DIC) need to be adapted:

1. replace dtTracerLev by PTRACERS_dTLev
2. replace THETA, SALT, etc. by their longstep averages from the table above. This should be made between #ifdef ALLOW_LONGSTEP ... #endif. (You need to #include "LONGSTEP.h")

The package is activated by including "longstep" in packages.conf. There is no "use_LONGSTEP", the package is always on when compiled.

The run-time parameters are set in data.longstep. The default is:

&LONGSTEP_PARM01 LS_nIter=1, LS_whenToSample=0, &

LS_nIter :: number of dynamical time steps between ptracer time steps. LS_whenToSample :: when to sample dynamical fields for the longstep average

0 - at beginning of timestep (reproduces offline results)
1 - after first THERMODYNAMICS but before DYNAMICS
    (use use old U,V,W for advection, but new T,S for GCHEM if
    staggerTimeStep=.FALSE.; reproduces online with
    staggerTimeStep=.FALSE. for LS_nIter=1)
2 - after DYNAMICS and second THERMODYNAMICS
    (use new U,V,W and T,S; reproduces online with
    staggerTimeStep=.TRUE. for LS_nIter=1)

Default is to sample dynamical fields at the beginning of the time step. This reproduces results from offline runs with unshifted time averages used for the dynamical fields.

To have a time stepping closer to the online model (and reproduce results for LS_nIter=1), use

LS_whenToSample=1 if staggerTimeStep=.FALSE. LS_whenToSample=2 if staggerTimeStep=.TRUE.