Running ADAS

DAS assimilation and forecast runs

Select DAS experiment to run

First, we choose a DAS run to simulate. For a list of runs

$ ls ~dao_ops

on NCCS discover. As an example, we choose the run d591_fpit (for more about GMAO products, please see https://gmao.gsfc.nasa.gov/products)

Look for ATAG in file ~dao_ops/d591_fpit/run/FVDAS_Run_Config

$ grep ATAG ~dao_ops/d591_fpit/run/FVDAS_Run_Config

setenv ATAG    GEOSadas-5_9_1_p7

This gives us the model tag.

Build model

$ cvs co -P -r GEOSadas-5_9_1_p7 -d g591p7 GEOSadas-5_9
$ cd g591p7/src
$ ./parallel_build.csh

Setup experiment (Run `fvsetup`)

Run fvsetup from the bin directory (g591p7/Linux/bin) with the following options (default value chosen for ones not mentioned): [For now we choose the default value for FVICS as well]

AGCM Resolution? [b72]
> C180

OGCM Resolution? [c34]
> f34

EXPID? [u000_C180]
> g591p7ops

OBSERVING SYSTEM CLASSES?
> ncep_prep_bufr

A pdf of the entire fvsetup session is provided here: File:Fvsetup session.pdf.

Fvhome

FVHOME is the home directory for fvDAS experiment. Resource files, restarts and system output will be stored under this directory. Usually it contains subdiretories:

ana/        first guess/analysis output
anasa/      job script to run stand-alone analysis
daotovs/    daotovs field output
diag/       diagnostic field output
etc/        listings and other odds & ends
fcst/       forecast run directory
fvInput/    process required inputs
obs/        post-analysis ODS files
prog/       prognostic field output
recycle/    latest restart files
rs/         restart files
run/        resource files

$FVHOME/.FVROOT gives the location of the model bin directory.

Some of these directories are stored in the archive. After running one cycle look in $ARCHIVE/g591p7ops (where g591p7ops is the experiment ID).

Move to the experiment directory

$ cd $NOBACKUP/g591p4_d72/run

At this point, one can edit CAP.rc.tmpl to run 12-hour experiment. Also, one can turn on MAPL timers and memutils

JOB_SGMT:     0 120000
NUM_SGMT:     1

MAPL_ENABLE_TIMERS: YES
MAPL_ENABLE_MEMUTILS: YES

To run one segment each of GSIsa.x and GEOSgcm.x add 'exit' before 'PART IV - Next PBS Job Segment' in g5das.j.

Data Assimilation experiment

$ cd $NOBACKUP/g591p4_d72/run
$ qsub g5das.j

g5das creates a scratch directory $NOBACKUP/fvwork.xxxx and runs the experiment there. After the initial setup, GSIsa.x is run with output written to ana.log followed by GEOSgcm.x which writes its output to stdout and fvpsas.log.