ticket_261_TicketSummary - ACCESS-NRI/accessdev-Trac-archive GitHub Wiki
Experimental setup
Overview
Two suites are used: a control (au-aa427) and a trial run without IASI observations (au-aa461). Both suites are based on APS2 ACCESS-G Rose suite, au-aa068 at ![1316].
-
Resolution: N512L70
-
PFM resolutions: N108/N216
-
Observation types used,
| obsgroup | platform | note | | --- | --- | --- | | | | | | screen | METAR (?), SYNOP | | | as | AMDAR's, AIREP's, TEMP, WIND, wind profilers | | | atovs | NOAA-15, NOAA-18, NOAA-19, MetOp-1, MetOp-2 | | | airs | Aqua | | | iasi | MetOp-1 and MetOp-2 | | | cris | Suomi-NPP | | | atms | Suomi-NPP | | | mtsatcsr | MTSAT-2 | | | amv | GOES-E, GOES-W, Aqua/MODIS, MTSAT-2, METEOSAT-7, METEOSAT-8, METEOSAT-9, METEOSAT-10 | ???? need to duble-check ????
For more details see here | | surface | SYNOP, METAR, ???? | | | gpsro | GRACE-A, COSMIC-1, -2, -3, -4, -5, -6, C/NOFS, GRAS-A, GRAS-B | For more details see here | | scat | ASCAT | | | ssmis | SSMIS | Note used |
-
Experiment period: 00Z 1 August 2015 until 18Z 31 October 2015
-
Spin-up period: 00Z 18 July 2015 until 18Z 31 July 2015
-
Forecast lengths: 12-hour steps up to 5 days (only 00Z-based forecasts are verified)
-
Analysis frequency: 00Z and 12Z
-
Fields to output: Vincent suggested T, q, u, v, Z, MSLP but from our runs instead of q relative humidity is written out
-
Levels: Vincent suggested 1000, 950, 900, 850, 800, 700, 600, 500, 400, 300, 250, 200, 150, 100, 70, 50, 30, 20, 10, 7, 5, 3, 2, 1 hPa but from our run 7, 5, 3, 2, 1 hPa are not available
-
Output in the Grib format on 0.5 deg x 0.5 deg grid
-
Aditional statistics:
- SatRad NetCDF writefiles for o-b statistics turned on
- o-a statistics from VAR not turned on
Note 1. ReconSST task fails because (I guess) somehow Cylc does not receive the completion of UM job. So I decided to use older Cylc and Rose versions under which this task worked. Following line was used to allow the use of old Cylc and Rose versions,
export CYLC_VERSION=6.4.1; export ROSE_VERSION=2015.04.1
NWP Configuration
Following summarises OPS control file versions used by UKMO (from collab wiki) and what I finally decided to use.
Stationlist
Time period | Version |
---|---|
20150717 18Z - 20150720 18Z | v396 |
20150721 00Z - 20150810 18Z | v397 |
20150811 00Z - 20150914 18Z | v398 |
20150915 00Z - 20151019 18Z | v400 |
20151020 00Z - 20151031 18Z | v401 |
Based on the the version used during the last time period - "ukmo/v401" - I modified it and came up with "users/au-aa427/aps2_r00.my00.v401". Some of the changes are,
- ATOVS_stlist.nl - changed thinning distance in extra-tropics
- IASI_stlist.nl - ditto
- MTSATClear_stlist.nl - commented out some stationlist keywords ('DayStart' and 'DayEnd') which were not available in OPS30.0.0
- Satwind_stlist.nl - commented out some stationlist keywords ('DayStart' and 'DayEnd') which were not available in OPS30.0.0
Screen stationlist
Time period | Version |
---|---|
20150717 18Z - 20150720 18Z | v126 |
20150721 00Z - 20150810 18Z | v127 |
20150811 00Z - 20150914 18Z | v128 |
20150915 00Z - 20151019 18Z | v129 |
20151020 00Z - 20151031 18Z | v130 |
Used the version which was used during the last time period - "ukmo/v130".
SatRad coeffs
Time period | Version |
---|---|
20150717 18Z - 20150820 18Z | v109 |
20150821 00Z - 20150823 18Z | v110 |
20150824 00Z - 20150921 18Z | v111 |
20150922 00Z - 20151004 18Z | v112 |
20151005 00Z - 20151020 18Z | v113 |
20151021 00Z - 20151031 18Z | v114 |
The version, "ukmo/v114" is what's used during the last period. Because I need to output SatRad NetCDF writefiles I put together "bom/users/au-aa427/aps2_r00_my00.v114" (based on "bom/users/au-aa094/aps2_r00_my00.v98").
RTTOV7 coeffs
Time period | Version |
---|---|
20150717 18Z - 20150820 18Z | v40 |
20150821 00Z - 20150823 18Z | v41 |
20150824 00Z - 20151031 18Z | v42 |
Used the version which was used during the last time period - "ukmo/v42"
MetDB Elements
Version | Note |
---|---|
v101 | bom/common/aps2.![00] is derived from this |
v102 | Added MVIRIClear_ESACSR.elements, changes for GPSRO TPD |
v103 | Fix typo in ATMS elements file |
v106 | Corrected GPSRO elements for TPD |
As the collab wiki page for MetDB elements files does not list dates I decided to use "bom/common/aps2.![00]" (this is based on "ukmo/v101")
N.B.
- There is a typo in ATMS_ATMS.elements (compare ATMS_ATMS.elements in v101 and v103) - check ATMS is processed correctly by OPS
- v103/GPSRO_GPSRO.elements was changed to include tangent point lat, lon and azimuth - check OPS processes GPSRO correctly
Scatwind coeffs
Time period | version used |
---|---|
20150717 18Z - 20150908 18Z | v24 |
20150909 00Z - present | v25 |
Used the version which was used during the last time period - "ukmo/v25"
Satwind coeffs
version used | note |
---|---|
20150717 18Z - 20151031 18Z | v19 |
Used the version which was used during the last time period - "ukmo/v19"
Sonde coeffs
Time period | version used |
---|---|
for a long time | v3 |
Used the version which was used for a long time - "ukmo/v3"
GPSRO coeffs
Time period | version used |
---|---|
latest version | v14 |
Used the lastest version - "ukmo/v14"
N.B.
- Is this O.K.?
Summary of versions of various OPS control files used=
OPS controlfile type | Version used | Comment |
---|---|---|
Stationlist | bom/common/aps2_r00.v401 | |
Screen stationlist | ukmo/v130 | |
SatRad coeffs | bom/users/au-aa427/aps2_r00_my00.v114 | To output SatRad NetCDF writefiles for o-b created suite-specific control file |
RTTOV7 coeffs | ukmo/v42 | |
MetDB Elements | bom/common/aps2.![00] | Need checking |
Scatwind coeffs | ukmo/v25 | |
Satwind coeffs | ukmo/v19 | |
Sonde coeffs | ukmo/v3 | |
GPSRO coeffs | ukmo/v14 | Need to check stdout log output |
set_ops_stationlist.ksh
_ ToDo Use stationlist file from au-aa094 as a starting point. You may need to copy stationlist files from Jasmin mirror to ~access/da as the files may need modification _
set_ops_screen_stationlist.ksh
set_ops_control_files.ksh
set_ops_bias_files.ksh
rose-suite.conf
Change CONF_START and CONF_STOP
ODB and OPS versions
-
ODB_VERSION: a. odb30.0.2
-
OPS version: a. OPS30.0.0 b. https://access-svn.nci.org.au/trac/ops/browser/branches/dev/satda/ops30.0.0 c. No changes are introduced to OPS code so locally generated AMV's are processed in the same as JMA-produced AMV's
Preparing Bufr files
Bufr files copied: 20150718 00Z till 20151031 18Z
Archiving
-
Model analysis and forecast files (in Grib) are archived to "raijin1:/short/dp9/jtl548/opendap". These were moved to MDSS,
jtl548/opendap
. The directory structure of the archive follows that of APS2 global:- control (with IASI) - jtl548/opendap/a427/yyyymmdd/hh00
- trial (without IASI) - jtl548/opendap/a461/yyyymmdd/hh00
-
SatRad writefiles are archived to "raijin1:/g/data/dp9/jtl548/glb"
-
deleted subdirectories,
au-aa427
andau-aa461
underraijin3:/short/dp9/jtl548/data/mdss/glb
Luke-warm start
-
"luke-warm" start files come from the BNOC APS2 SCS trial suite, "saneg". The suite's archive location is,
ngamaidm04:/g/ns/lustre/users/bnoc/rto/access_g/saneg/2015_TEST
-
start dump (iau-start-dump) from 20150717 12Z
-
analysis increment (an_inc) from 20150717 18Z
-
soil moisture (an_smc) from 20150717 18Z
-
bgerr (bgerr.new) from 20150717 18Z
-
all the warm-running files are from pre-operational APS2 ACCESS-G, saneg.
Bias coefficient updates
RUN 1: Spin-up for bias correction
-
Start date: 20150717 18Z (luke-warm start; first full OPS, VAR and UM were for the cycle time of 20150718 00Z)
-
End date: 20150731 18Z
Note 1. Initial bias regression coefficients file came from au-aa094 and the file is quite old
Note 2. After the cycle, 20150731 18Z I attempted to update the bias coefficients. However, there was a problem with NOAA-15 as the script, "LaunchBiasCorrection.ksh" did not contain NOAA-15 in the list of the satellites for ATOVS. The script should be updated!
RUN 2: After first update of bias regresssion coefficients
- Start date: 20150801 00Z
- End date: 20150831 18Z
RUN 3: After second update of bias regresssion coefficients
- Start date: 20150901 00Z
- End date: 20150930 18Z
RUN 4: After third update of bias regresssion coefficients
- Start date: 20151001 00Z
- End date: 20151031 18Z
Diary
au-aa427
Cycle time | Failed task | Reason for failure | Action taken |
---|---|---|---|
2015071718 | get_mtsatcsr | bufr2odb failed with stdout, |
BUFR TABLES TO BE LOADED B0000000000000010000.TXT,D0000000000000010000.TXT BUPMRK : INCREMENT VALUE FOR COMPRESSED DATA TOO BIG. MY_EXIT-code = 2 | reset to succeeded | | 2015072012 | get_mtsatcsr | Task failed with stdout,
BUFR TABLES TO BE LOADED B0000000000000010000.TXT,D0000000000000010000.TXT BUPMRK : INCREMENT VALUE FOR COMPRESSED DATA TOO BIG. MY_EXIT-code = 2 | reset to succeeded | | 2015080206 | Analysis108 | failure at early step (?) | rerun task | | 2015080312 | Analysis216 | Last VAR routine called before failure is Var_InitMatchCovKE.f90 | rerun task | | 2015080412 | Analysis216 | Last VAR routine called before failure is Var_InitMatchCovKE.f90 | rerun task | | 2015080718 | Analysis216 | Last VAR routine called before failure is Var_InitMatchCovKE.f90 | rerun task | | 2015081818 | Ops_surface | Problem reading ODB? | Decided not to assimilate surface observations for this cycle (same should happen with au-aa461). Reset to succeeded | | 2015090312 | Ops_mtsatcsr | Size of Bufr file zero | reset to succeeded | | 2015090400 | Ops_mtsatcsr | Size of Bufr file zero | reset to succeeded | | 2015090406 | Ops_mtsatcsr | Size of Bufr file zero | reset to succeeded | | 2015090412 | Ops_mtsatcsr | Size of Bufr file zero | reset to succeeded | | 2015092900 | get_amv | Bufr files for AVM's are all of zero size | reset to succeeded | | 2015092900 | Ops_amv | Bufr files for AVM's are all of zero size | reset to succeeded | | 2015100100 | rose suite-run | When starting the suite made mistake and had to interrupt 'rose suite-run'; this resulted in log files not saved and lost | reran 'rose suite-run' |
au-aa461
Cycle time | Failed task | Reason for failure | Action taken |
---|---|---|---|
2015071718 | get_mtsatcsr | bufr2odb failed with stdout, |
BUFR TABLES TO BE LOADED B0000000000000010000.TXT,D0000000000000010000.TXT BUPMRK : INCREMENT VALUE FOR COMPRESSED DATA TOO BIG. MY_EXIT-code = 2 | reset to succeeded | | 2015072012 | get_mtsatcsr | bufr2odb failed with stdout,
BUFR TABLES TO BE LOADED B0000000000000010000.TXT,D0000000000000010000.TXT BUPMRK : INCREMENT VALUE FOR COMPRESSED DATA TOO BIG. MY_EXIT-code = 2 | reset to succeeded | | 2015080206 | Analysis108 | failure at early step (?) | rerun task | | 2015080412 | FC and FC_short | failure during IAU step: last UM routine called is mpl_allreduce.f90 | rerun task | | 2015080500 | Analysis216 | Last VAR routine called before failure is Var_InitMatchCovKE.f90 | rerun task | | 2015081400 | ReconSST | FailureOfReconSST | rerun task | | 2015081818 | Ops_surface | Problem reading ODB? | Decided not to assimilate surface observations for this cycle (same as what happened with au-aa427). Reset to succeeded | | 2015090312 | Ops_mtsatcsr | Size of Bufr file zero | reset to succeeded | | 2015090400 | Ops_mtsatcsr | Size of Bufr file zero | reset to succeeded | | 2015090406 | Ops_mtsatcsr | Size of Bufr file zero | reset to succeeded | | 2015090412 | Ops_mtsatcsr | Size of Bufr file zero | reset to succeeded | | 2015092900 | get_amv | Bufr files for AVM's are all of zero size | reset to succeeded | | 2015092900 | Ops_amv | Bufr files for AVM's are all of zero size | reset to succeeded | | 2015102012 | [OPS] family and other subsequent families | Some families seem to be waiting for pre-requites to finish and some pre-requites are not visible on gcylc. To avoid having to do manual triggering decided to warm-start from 2015102012 since the cycle 2015102006 seems to have finished successfully | warm-start from 2015102012 |