Data written to the history files are maintained in a separate buffer distinct from the main model buffers. This
buffer
may
be
cycled
to
the SSD (unit sunit) or maintained in-core. In either case its existence must
span
several
time
levels because most fields written to the history file are time-averaged. Therefore, the data must be carried across
time
levels
between
the outfld calls which accumulate them into the history buffer. The history
buffer
is
maintained
as a singly-dimensioned array, with integer pointers delimiting locations of the various fields within the history
buffer
array.
Refer
to
"Details of CCM3 Input and Output Datasets" under "History Buffer" for
more
information.
Questions on these pages can be sent to... erik@ucar.edu .