...
write is mainly used to save data from fieldsets, a temporary storage, into files, see Fieldsets.
Fieldsets
A fieldset is a temporary storage for fields, either retrieved from MARS or read from a UNIX file. Fieldsets can be seen as variables in a MARS request, and they can be referenced in subsequent MARS requests in the same execution. This is an example how to retrieve data into a fieldset named analysis:
retrieve,
class = od,
type = analysis,
stream = oper,
expver = 1,
date = -1,
time = 12,
levtype = pressure levels,
levelist = 1000/850/500,
param = temperature,
fieldset = analysis
|
At the end of the execution, all fieldsets are released. Therefore, data in unsaved fieldsets will be lost. Fieldsets can be saved as UNIX files with write requests, like in this example:
write,
fieldset = analysis,
target = "data"
|
Note that fieldsets are in fact using temporary disk storage on the host where MARS is executed, usually in $TMPDIR directory. When dealing with big fieldsets, you are advised to make such variable point to a suitable directory where temporary data can be stored (e.g. $SCRATCH on ecgate/HPC).
One can extract fields (sub fieldsets) from a fieldset using square brackets. If X is a fieldset, X[2] is the second field, X[1,10] comprises the first 10 fields, and X[2,8,2] comprises every second field of x starting from 2 up to 8keyword fieldsets.
Multiple requests in one call to MARS
...