-----------------------------------------------------------------
Current history file:

lev = 29: vertical dimension of interfaces

lev(lev): vertical coord array at interfaces (zp = -7 to 7 by 0.5)

all fields: (lon,lat,lev,time)

There are at least two problems with the current history files:

1) The model calculates some (most) fields at midpoints, and a few
   fields at interfaces, however, the history provides only one
   vertical coordinate array, at interfaces. This means fields
   calculated on the midpoints are presented incorrectly on the
   history.

2) The bottom interface level (zp -7.0) for T,U,V are calculated
   separately in the model. This is ok, but they are then stored
   in the top slot (k=nlevp1, zp=7.25) on the history. This is not
   obvious and is confusing for anybody not familiar with the
   histories. 

-----------------------------------------------------------------
Proposed new history file:

Goals:
  1) Accurately represent the model results

  2) Solve the 2 problems with current history files described above

  3) Decouple model and post-processors, i.e., make it relatively
     easy for an outsider to understand and read the history files
     without having to use our elaborate post-processors.

  4) Conform, at least in part, with the netCDF CF conventions.

Provide two coordinate arrays (both dimensioned 29):

lev = 29; vertical dimension of fields at midpoints
lev(lev); 
  lev: long_name = "level midpoints"      (e.g., lev = -6.75 to 7.25 by 0.5)
  lev: units = "ln(p0/p)"
  lev: positive = "up"

levi = 29:  vertical dimension of interfaces
levi(levi):
  levi: long_name = "level interfaces"    (e.g., lev = -7 to 7 by 0.5)
  levi: units = "ln(p0/p)"
  levi: positive = "up"

TN,UN,VN,O2,O,N4S,NO,N2D,OP,O2P,TE,TI: (lon,lat,lev,time)

NE,W,Z,POTEN: (lon,lat,levi,time)

New variables: 
  double press(lev);    
    press:long_name = "Pressure at midpoints"
    press:units = "mb"
    press:positive = "up"
  double pressi(levi):
    pressi:long_name = "Pressure at interfaces"
    pressi:units = "mb"
    press:positive = "up"

  double tlbc(lon,lat); 
    tlbc:long_name = "Bottom interface boundary of TN"
  double ulbc(lon,lat); 
    ulbc:long_name = "Bottom interface boundary of UN"
  double vlbc(lon,lat);
    vlbc:long_name = "Bottom interface boundary of VN"

  dday(time):    Decimal day (alternate time coord array)

New Field Attributes:
  cell_method = "zp: interfaces"
                "zp: midpoints"
                "zp: unknown"

-----------------------------------------------------------------
Proposed new software:

Provide short, concise procedures to read a generic TGCM history file, 
in fortran, IDL, and NCL. These should be written so that they are
easy to import into a user's own visualization and analysis software.

Provide a command (based on above procedures) to read and print
fields from a history file (various options such as interfaces vs
midpoints, which variables, slices, etc).