force/README.force:

Interpolation of Naval Research Laboratory (NRL) format wind/flux files
on their "native" grids to HYCOM model grid.

../../ALL/bin/wind_stat		lists contents of "native" file
../../ALL/bin/wind_stat.1	man page for wind_stat
../../ALL/force/src/		softlink to latest source directory
force/coads/			COADS winds/fluxes on HYCOM grid
force/offset/			offset field required by wind interpolation
force/plot/			plot net radiation flux
force/new_force.com		edits scripts for a new region

Before interpolating COADS winds to the hycom grid, an offset file 
is required.  In this case it is all zeros.  See force/offset.

The interpolation program (in hycom/ALL/force/src) reads (say) COADS 
wind/flux files in NRL format (a Fortran unformatted sequential file with 
a single header record identifying wind dates, followed by the wind/flux 
data).  COADS in the required format is available from the HYCOM ftp site, 
see: ftp://obelix.rsmas.miami.edu/awall/hycom/coads_for_hycom.tar.gz
The hycom/ALL/bin/wind_stat command lists the contents of a native wind
or flux file.  There is also a wind "offset" input file, that allows the
annual mean wind to be from a different wind data set.  Usually the offset
is zero, but it must always be read in and so the typical first step for
wind generation is to create force/offset/tau[en]wd_zero.[ab] for the 
specified model region.  The offset can also be a different field for each 
sample time.  This allows combining a climatology with an anomaly field.

The output data sets are atmospheric forcing in MKS, with heat flux positive
into the ocean (see README/README.src.MICOMvsHYCOM).  They also include any
bias or minimum wind speed constraints.  For example, in force/coads, 
compare coads_mon_flux.com (zero bias) to coads_mon_flux+070w.com (70w bias).  
An all zero precip file input to HYCOM indicates that there should be no e-p
surface salinity forcing, see force/coads/precip_zero*.  By default the 
output wind stress components are on the "native" u and v grids, but setting
NAMELIST variable IGRID=2 will output wind stress component on the pressure
grid (which is always used for all other atmospheric forcing fields).
When running HYCOM, use wndflg=1 for u/v winds and wndflg=2 for winds on
the pressure grid (wndflg set in blkdat.input).  Note that in all cases
the HYCOM wind stresses are orientated w.r.t. the local HYCOM grid, i.e.
taunwd (tau_y) is only actually "North-ward" when the HYCOM grid is E-W/N-S.  
The actual surface forcing fields and their units are:

   surface wind stress,      tau_x,  units are N/sq_m (positive  eastwards)
   surface wind stress,      tau_y,  units are N/sq_m (positive northwards)
   wind speed at 10 m,       wndspd, units are m/s
   surface air temperature,  airtmp, units are degC
   water vapor mixing ratio, vapmix, units are kg/kg
   precipitation,            precip, units are m/s    (postive into ocean)
   total     radiation flux, radflx, units are w/sq_m (postive into ocean)
   shortwave radiation flux, swflx,  units are w/sq_m (postive into ocean)

The same interpolation programs can be used as part of the standard HYCOM
run script to produce the 12-hrly or 6hrly winds/fluxes for actual
calendar days from operational center wind/flux files on their native grid 
in NRL format.  This is done "just in time", i.e. the files for the NEXT
model segment are generated while the CURRENT model segment is running.

See also:

force/coads/README.force.coads
force/plot/README.force.plot
ALL/force/src/README.ALL.force.src