The data files themselves should have the following naming convention (for ease of processing by the ICC):
| Snapshot Variables | Syymmdd.xxx |
| Decad-Mean Variables | Dyymmdd.xxx |
| Fixed Fields | FIXED.xxx |
| Monthly-Mean Diurnal Cycles at Specific Points | Pyymm.xxx |
The data should be in the order and units described in the GSWP
document. The structure of the data should be as follows:
Only the "control" integration needs to be submitted to the ICC.
The data should be submitted on an exabyte (8mm) tape.
There should be 4 sets of data per 2-year integration. They are
in order (with numbering from the documentation):
2.6.1 a) Snapshot variables each decad
b) Mean variables each decad
c) fixed fields
2.6.2 a) Monthly mean diurnal cycles at selected points
The first set on the tape should be 2 years worth of decad snapshot
variables. There are 73 time periods which should be written to tape,
the initial (post spin-up) state (0000UTC01Jan1987) and the values from
0000UTC on the 1st, 11th, and 21st of each month through the last time
(0000UTC01Jan1989). There are 35 snapshot variables listed in the
document:
moisture production variables (5)
forcing variables (6)
state variables (24 accounting for 5 layers for 3 of the variables)
The second set on the tape should be 2 years worth of decad-mean
variables. There are 72 time periods, 3 per month, with averaging
periods from 0000UTC01-2359UTC10, 0000UTC11-2359UTC20, and
0000UTC21-2359UTC of the last day of the month (remember that Feb1988
has 29 days!!). There are 30 decad-mean variables listed in the
document:
production variables (7)
water balance terms (9)
heat balance (7)
diagnostics (3)
residuals (4)
The third set contains a single time period and contains the time-
invariant model dependent soil proerties necessary for soil moisture
conversions. There are 6 fields listed in the document:
fixed fields (6)
The last set contains site specific data -- mean monthly diurnal
cycles for selected points. There are 24 time periods (each calendar
month). There are 14 variables listed in the document:
state variables (4)
heat balance (7)
diagnostics (3)
There are 13 points to include. The data should be saved as a vector
of 13 (as opposed to the vector of 14637 for global land points above).
The points are listed in Table 7, p.37 of the GSWP document. The 13th
point not included in table 7 is:
EDE-Holland at 52.02N, 5.45E for 1988.
NOTE: At this time there is no intention for producing the hourly variables
for specific sites from the 1987-1988 data (2.6.2 b)). This will be of more use
when the ISLSCP Initiative II data become available, covering more of
the periods of the major field experiments.
For each output variable and at each output time step, the two dimensional
fields should have a header in character format, then binary data in IEEE
format.
Sample fortran code for one field (2 records):
C==============================================================
CHARACTER*64 HEADER( 10 )
REAL GDATA (14637)
C