The text fields which make up a file name are
I
(for instantaneous data), D
(indicating data averaged over days),
M
(indicating data averaged over months), or Y
(indicating
data averaged over years). If the rest of the field is numeric,
then it may be of the form 'yymmdd' or 'yymmddhh', where the y's stand for
year digits, the m's for month number digits, d's for the digits of the day of
the month, and h's for the hour of the day. If the part of the
date field past the leading character is
alphabetic, then it is a three-digit base-26 number (A
=0, Z
=25)
indicating the number of days since 1 January 1970.
T
. If the rest
of the field is numeric, then it is of the form 'mmmmss', where
the m's represent minutes, and the s's represent seconds.
X
. If there are no more
characters in this field, or if only one character,
P
(indicating that the file contains packed data)
follows, then the file uses XDR
data formats. Otherwise, the lead
character is followed either or both of two subfields:
a C
followed by either A
or E
, indicating whether
character data is ASCII or EBCDIC; or
a B
followed by a floating-point format
indicator (E
for IEEE, V
for VAX, I
for IBM mainframe,
and Y
for Cray), a byte-order flag (B
for
big-endian, L
for little-endian), a word-order flag (B
for
big-endian, L
for little-endian), record-header flag
(C
for C-style stream files without record
headers, F
for Fortran record headers), a
flag indicating the type of file record structure (S
for pure
stream files, V
for variable-length records, and
F
for fixed-length records), and a packing flag
(P
for packed data, U
for unpacked).
E
.
G
, and the characters following
should follow some local convention to indicate
grid type, location, grid spacing, etc. If needed,
decimal points may be represented by Level-2 delimiter
characters (see Section B.3.2).
F
.
S
.