!FAMOUS Standard Versions
FAMOUS standard versions are identified by the UM runid of a control run on some platform. We recommend identifying versions of the model by the runid of a standard version to which they are scientifically equivalent. Equivalence is hard to define. For [[HPCx|porting to HPCx]], we used statistical tests made with IDL programs (available on request) on a number of fields from 100-year control runs. Porting to a new platform often introduces scientific differences, typically because of compiler-dependency, optimisation or unsafe code, so scientific equivalence must be checked carefully.
Note that single-PE operation is not supported by these versions. Use MPP even when running on a single processor.
For technical updates and differences between standard jobs see [[HPCxJobUpdates|standard job updates]].
Where bugs are found post-release, they will be flagged here. In general, fixes will be included in the next release but not backported into older jobs. Bugs flagged up for any given version can be assumed to exist in all previous versions. More information on these issues and their fixes can usually be found on [[CurrentWorkRobin|Robin's]] page.
----
'''LATEST''' (Robin's jobs, not ''strictly'' release versions yet - ie. not available from Annette. Job names may change)
%green%xfhcy%%: MOSES 2.2 tuning job (FAMOUS*).
This replaces xdkub as the release candidate for the MOSES2.2 version of FAMOUS. It's currently being tuned by Jonny Williams in Bristol. In addition to MOSES2.2 - and hence a land-carbon cycle - this version also contains:
* top-level friction to fix the unphysical winds "properly" - improves atmospheric vertical structure and cuts down on the over-enthusiastic northern hemisphere filtering, which improves atmospheric heat transport and the winter cold bias. This also lets you have more realistic ozone concentrations, and return to the better-justified [=HadCM3=] sea-ice albedo
* sea-ice has velocity independent of the surface ocean, can be forced more realistically by the winds
'''known bugs:'''
-> diagnostics of ocean vertical motion seem to be currently inaccurate
----
%green%xfxwb%%: [=EMICmip=] (AR5) control run.
This is basically a bug-fixed version of the xdbub release described in the 2008 GMD paper (see below). The GMD FAMOUS paper will be updated to reflect this version shortly.
'''features'''
* Disappearing snow on sea-ice at coastal points, the interchange of ocean sensible and latent heat fluxes in the boundary layer stability calculation have been corrected and the ocean solar absorption ''without'' [=HadOCC=] are all fixed
* A trivial cap on the unphysical top level winds makes the model far more stable under perturbed climate conditions
* UKMO "FHL bugfix" mods backported from vn6.1 - all minor.
* Iceberg calving flux recalibrated for the better spun-up climate to reduce global salinity drift
* (final version will also have mods to fix the zooplankton diagnostic bug and allow updating of land-ancils)
'''known bugs:'''
-> diagnostics of ocean vertical motion seem to be currently inaccurate
----
'''OLDER'''
%green%xdkub%%: MOSES 2.2 tuning job.
'''known bugs:'''
-> The climate is considered not suitable for scientific use. This job has been released to allow users to do work themselves to improve the climate.
-> MEAD diagnostics don't work @ Bristol or for the pgi compiler on [=HECToR=]
-> Zooplankton production diagnostic actually shows export calcite
Jobs for supported platforms can be found under owner @@annette@@ on the puma UMUI:
* xdkub: quest-hpc
* xdkuc: [=HECToR=] pathscale compiler (Note that this job crashes with a negative theta value at timestep 213)
* xdkud: [=HECToR=] PGI compiler
----
%green%xdbub%%: Includes [=UTOPIA=] and [=HadOCC=]. This job has the same physical climate as xcpsa below, but much improved marine biogeochemistry from the [=HadOCC=] component. [=HadOCC=] tracers are now advected with the 3rd order flux-limited UTOPIA scheme, and air-sea CO2 fluxes are much better. This release corresponds to the model write up published in [[http://www.geosci-model-dev.net/1/53/2008/gmd-1-53-2008.html|Geoscientific Model Development 1, 53-68, 2008]]. A few extra bits of climate analysis can be found on the [[xdbub_extra]] page.
'''known bugs:'''
-> Snow on land at coastal points is overwritten by the snow lying on the sea-ice fraction of that point. Implies a warm bias around Greenland and Antarctica.
-> Surface sensible heat and moisture fluxes are swapped at the end of SFFLUX5A. Doesn't seem to have a large influence on the climate - causes a small northern hemisphere warming in winter?
-> Running with [=HadOCC=] off does undefined things with the solar absorption in the ocean surface. Doesn't seem to have a serious climate effect on the platforms it has run on.
-> Updating land surface ancils during a run is broken.
-> MEAD diagnostics don't work @ Bristol or for the pgi compiler on [=HECToR=].
-> Bounds checking error in PPHEAD1A (not related to LBLEV issue) - a [=STASHmaster=] issue, we think.
-> Zooplankton production diagnostic actually shows export calcite
Standard jobs for supported platforms can be found under owner @@annette@@ on the puma UMUI, and will remain fixed.
* xdbua: [=HPCx=]
* xdbub: quest-hpc
* xdbuc: [=HECToR=] pathscale compiler
* xdbud: [=HECToR=] PGI compiler
There is also a standard job with [=HadOCC=] off:
* xdbun: [=HECToR=] pathscale compiler
All FAMOUS input data, mods, [=STASHmasters=] and post-processing scripts are stored on puma in @@/home/famous/@@.
----
%green%xcpsa%%: Incorporates a number of climate tunings and additional features. Most notably, the high northern latitude cold bias of previous versions is much improved. Plots and more detailed documentation can be found here: [[http://cgam.nerc.ac.uk/pmwiki/FAMOUS/index.php/Main/Dec07]]. A few extra bits of climate analysis relevant to this model can be found on the [[xdbub_extra]] page.
'''known bugs:'''
-> Running with [=HadOCC=] off (the released configuration) does undefined things with the solar absorption in the ocean surface. Doesn't seem to have a serious climate effect on the platforms it has run on.
-> Salinity correction needs tweaking for guaranteed correct performance on all PE configurations
-> MEAD diagnostics don't work @ Bristol
"Standard jobs" for supported platforms. These can be found under the username annette on the puma umui, and will be kept up to date with technical fixes.
* xcpsa - [=HPCx=]
* xcpsb - quest
FAMOUS files are now stored on NCAS's UM server, [[puma->http://ncas-cms.nerc.ac.uk/content/view/736/187/]], under the directory @@/home/famous@@. We would encourage FAMOUS users to register for this service. For more information, contact [[NCAS-CMS->http://ncas-cms.nerc.ac.uk]].
HPCx (xbyva=xcpsa) and QUEST (xbyvq=xcpsb) results have been statistically compared as for the adtan comparisons below. Download the [[results of statistical comparison->http://www.famous.ac.uk/Stats/QUEST_tarograd.txt]] and [[accompanying plots->http://www.famous.ac.uk/Stats/QUEST_tarograd.ps]]. As noted in the [[Dec07|caveats]], there appears to be a small salinity drift issue on QUEST that can be seen in these comparison plots.
----
xcggb: New version of salinity correction - two minor bugfixes, correction also applied to the surface virtual fluxes of TCO2 and Alkalinity due to P-E+R (if [=HadOCC=] is switched on). The corrections are now [=STASHed=] in units of <tracer units> per second, rather than <tracer units> per timestep, so restarts that used the old scheme will have screwy corrections for the first year.
'''known bugs:'''
-> Salinity correction still not ''quite'' right. All of the changes to the salinity correction mod have very small effects, and it's difficult to tell as the model as a whole doesn't perfectly conserve water anyway.
-> MEAD diagnostics don't work @ Bristol
'''Download''' (14/06/07)
* [[README.txt->http://www.famous.ac.uk/StandardVersions/xcggb/README.txt]]
* [[famous_umui.tar.gz->http://www.famous.ac.uk/StandardVersions/xcggb/famous_umui.tar.gz]]
* [[famous_mods.tar.gz->http://www.famous.ac.uk/StandardVersions/xcggb/famous_mods.tar.gz]]
* [[famous_data.tar.gz->http://www.famous.ac.uk/StandardVersions/xcggb/famous_data.tar.gz]]
Tested on quest cluster (Bristol).
* Job-id is xcggc under puma umui.
* Results found to be statistically compatible with an equivalent 100 year run of xcggb on [=HPCx=]. Download the [[results of statistical comparison->http://www.famous.ac.uk/Stats/tarograd_xcggb_xcggc.txt]] and [[accompanying plots->http://www.famous.ac.uk/Stats/tarograd_xcggb_xcggc.ps]].
* The FAMOUS files are installed under @@[=~um/FAMOUS64=]@@.
----
xbyvs: adtan with salinity drift corrected. (Standard job for HPCx is xcgga in the Reading UMUI.) Model largely as adtan (Jones et al., 2005) but includes a volume-uniform time-dependent correction to conserve water in the open oceans to cancel the drift from the rigid-lid salt-flux representation of ocean surface water fluxes. Isolated inland seas are not affected by this fix: their runaway salinities due to local P-E+R imbalance are simply capped at 0 or 50.
'''known bugs:'''
-> The salinity correction mod isn't quite right, and xbyvs is now not recommended for use (see xcggb above).[[CurrentWorkSalinitydrift|Further discussion]]
[[ xbyvs_valnote | Validation note]]
'''Download''' (23/03/07)
* [[README.txt->http://www.famous.ac.uk/StandardVersions/xbyvs/README.txt]]
* [[famous_umui.tar.gz->http://www.famous.ac.uk/StandardVersions/xbyvs/famous_umui.tar.gz]]
* [[famous_mods.tar.gz->http://www.famous.ac.uk/StandardVersions/xbyvs/famous_mods.tar.gz]]
* [[famous_data.tar.gz->http://www.famous.ac.uk/StandardVersions/xbyvs/famous_data.tar.gz]]
Tested on ormen (Bristol).
* Results found to be statistically compatible with an equivalent run on HPCx.
* Download the [[results of statistical comparison->http://www.cgam.nerc.ac.uk/~annette/famous/tarograd_ormen.txt]] and [[accompanying plots->http://www.cgam.nerc.ac.uk/~annette/famous/tarograd_ormen.ps]].
* The FAMOUS files are installed under /home/moxsw/famous.
----
adtan: The phase 5 tuning of [[Jones et al. (2005)->http://www.famous.ac.uk/Papers/jones05famoustuning.pdf]]
'''known bugs:'''
-> there is a bug in the Mead diagnostics. - '''NB this is now "fixed". If meadlengths.mod is still enabled in your version, turn it off'''
-> LBLEV may not be set correctly in ocean 3D fields (this is OK on HPCx but not our Linux platform).
-> packing is switched off in PP output because some problems were found in ocean temperature fields as a result of the combination of packing and the presence of very large salinity values in inland seas.
'''Download''' (15/01/07)
* [[README.txt->http://www.famous.ac.uk/StandardVersions/adtan/README.txt]]
* [[famous_umui.tar.gz->http://www.famous.ac.uk/StandardVersions/adtan/famous_umui.tar.gz]]
* [[famous_mods.tar.gz->http://www.famous.ac.uk/StandardVersions/adtan/famous_mods.tar.gz]]
* [[famous_data.tar.gz->http://www.famous.ac.uk/StandardVersions/adtan/famous_data.tar.gz]]
Note that
* various diagnostics (3236 3237 3360 3361 3364 3365) have been given a modified space code in the stashmaster file so that they are always present in the D1 array even if not requested by the user for output. This is because the coastal tiling code uses them as workspace. Probably this will not be necessary with MOSES 2.2.
* the filtering safety factor has been increased, to reduce the incidence of dynamical crashes due to negative pressure.