FAMOUS

Main.MosesCoastalTiling History

Hide minor edits - Show changes to output

Changed lines 1-270 from:
''Current status: Testing 4.5 job with Moses2.2 included.''


!Testing


!!Current job


'''xcddj'''
* Based on standard [=HPCx=] [=HadAM3=] job.
* Completes 30 day run.
* With Moses2.2 set up as advised in [=HCTN30=], plus reconfiguration, extra ancillary fields and extra fixes.
* Note: not selected
** amv2f406 (radiative canopy model)
** prognostic snow albedo (spectral albedos)


!!Tests


* Restarts - '''xcddn'''
## Run 3 days in 1 3 day chunk
## Run 3 days in 3 1 day chunks
-> Note 1 dump period in both cases.
-> Compare results with cumf

* Number of processors
## Run for set time on 16 processors
## Repeat for 32 processors
-> Compare output files

* Turn on bounds checking

* Tiling
## Run with 9 tiles and "standard" vegetation fraction file
## Run with 9 tiles averaged down to 1 (i.e. NTILES=1 in namelists)
## Run with 9 tiles all the same - the same veg type & fraction
-> Look at difference in results.
-> Expect similarities(?)

* Set up specific STASH and output
## Run job with Moses2.2 switched off (i.e. standard Moses1)
## Run job with Moses2.2
-> Compare climatologies


!Action Plan


''AIM: To implement Moses2.2 with coastal tiling in FAMOUS.''


!!Summary (Lois)


# The 4.5 and 6.1 code tree diagrams show that there have been considerable code rearrangements of the boundary layer code.
** => Back-fitting 6.1 code may be problematic.

# The difficulty in doing a straightforward difference between 6.1 .f files and 4.5 .f files show that there has been a change in coding styles from 4.5 to 6.1
** => Back-fitting 6.1 code may be difficult


!!Strategy


# Start with [=Moses2.1=] and upgrade this to [=Moses2.2=], i.e. put in mods from [=HCTN30=] for vn4.5.
** [=HadAM3=] with basic 2.2 mods
** Set up ancillary files
** Design a testing mechanism

# Upgrade this [=Moses2.2=] basic version to as near 6.1 as possible. This will be difficult to judge - it will have to be an iterative procedure.
** Build the "best possible" [=Moses2.2=] at N48
** Try FAMOUS at N24

# Put in coastal tiling mods which are only needed when not using N48 and standard ocean.
** Test this with FAMOUS
** Compare and contrast FAMOUS climate with & without Moses2.2


!Background


!!Versions of Moses2.2


There are versions of Moses2.2 at 4.5 (without coastal tiling), at 6.1 (with coastal tiling), and as stand-alone JULES.


!!!Summary of development


The original version of Moses2.2 was a set of mods to 4.5. These were not formally added to the UM until around 5.2 (new dynamics). Coastal tiling was added to the UM at version 5.3 (there are mods for 5.2). There have not been many changes since (extra functionality, possible scientific improvements?) and the latest incarnation, JULES, is essentially Moses2.2 at 6.1. Most importantly, ''the structure of JULES is the same as in the 4.5 mods''.


!!!4.5 mods


To run Moses2.2 in 4.5 the following options must be selected in the UMUI:

Atmosphere -> Section by section choices
* Boundary layer (Section 3): MOSES-II (7A)
* Surface Hydrology (Section 8): MOSES-II (7A)

Moses2.2 is implemented in 4.5 as a set of mod files, of which we currently have three sets:
# The original set of mods, as specified in the Moses2.2 documentation ([=HCTN30=]).
# Dave Lawrence's mods from Tom Osborne's job '''xbhva'''
# A "clean" set from Martin Best

Details of decks added/changed etc in @@/home/annette/moses/compare/moses_mods.xls@@


!!!!Standard mods from [=HCTN30=]
[@
amv1f406
apa1f406
are1f406
are2f406
are3f406
newdecks

arerf406 - reconfiguration
amv2f406 - radiative canopy model

In directory /home/annette/moses/compare/standard_mods except arerf406.
@]



!!!!Dave Lawrence's mods
[@
apa1f406 - As HCTN30 standard
are1f406
are3f406

abx0f406 - Fixes
abx1f406
abx2f406
abx3f406
abx4f406

amv1f406.xdiag.dml - Modified to add passing of soil moisture to TILE_ALBEDO
are2f406.xdiag.dml
newdecks.xdiag.dml

can.capacity.mod - Define heat capacity by canopy height instead of LAI
can.heatcap.mod - Select various science
kill.phenol.mod - Remove call to phenology
vegparam.dml - Modified plant functional type params
vegscatter.mod - Correct scattering of updated veg fields

All in  directory /home/annette/moses/compare/xbhva_mods.
@]


!!!!Martin Best's mods
[@
apa1f406 - As HCTN30 standard
are1f406
are2f406
are3f406
newdecks

amv1f406_bh - Add Beljaars & Holsten stable stability functions in PHI_M_H
charnock_8A - Alternative computation of Charnock formula
decks8A - Add A03_8A to ifdefs
z0mod - Change roughness data

All in directory /home/annette/moses/compare/martinbest_mods.
@]


!!!6.1
 

Moses2.2 with coastal tiling is included as part of the UM at version 6.1. An example job is the standard IPCC run '''xcdaa''' which has the following section choices:

Atmosphere -> Section by section choices
* Boundary layer (Section 3): MOSES-II (8B)
* Surface Hydrology (Section 8): MOSES-II (7A)
* Vegetation (Section 19): Prescribed (1A)


!!!JULES


JULES is a stand-alone model, see [[http://www.jchmr.org/jules/index.html | JULES web site.]]

The current release, version1.0, was officially launched at the workshop at Reading University on 2-3 October 2006. There first version was taken from Moses2.2 at 6.1. The plan is to include JULES back in the UM within the next 12 months.

There is also a gridded version of JULES, which is the same as the standard code except that it allows users to specify longitude and latitude grids and run JULES over fixed areas. In this way, calculations can be run in parallel over several processors.

CEH Development plans:
* Modularisation, coupling interfaces for ESMs, code maintenance
* Improvement of hydrological and soil biogeochemical processes
* Incorporation of interactive nitrogen cycling

JULES is designed to be a community model, with a 2 year release-cycle. During this period, changes or developments made by the community can be submitted to the JULES developers for formal inclusion in the model. Decisions about inclusion will be made by the management team. In a similar way to the UM, changes will not replace previous code;  there will be various switches for section choices and some kind of version management system.


!!Coastal tiling


FAMOUS currently uses Moses1 with coastal tiling mods, but, as far as we are aware, there are no coastal tiling mods for Moses2.2 at 4.5.

Coastal tiling was formally added to the UM at 5.3 with Moses2.2. There are mods for coastal tiling with Moses1 at 4.5 (used by FAMOUS), and Moses2.2 at 5.2.


!!!5.3 mods


Nic Gedney has a list of 5.2 mods required for Moses2.2 coastal tiling in @@/home/annette/moses/coastal_tiling/documentation/5.3_042002.ps@@


!!!!Reconfiguration mods
[@
ang0f503r.mf77 - ancill related mods
ang0h503.mh - ancill related mods
ang0n503.mf90 - ancill related mods + 5.1 tstar bug fix (tstar_fix.mf90
buffix.mf90 - bug fix in 5.1 reconfiguration code
@]

!!!!Model mods
[@
ang0f503.mf77 - sfc/bdy layer mods
ang0h503.mh - ancill related mods
ang0f503r.mf77 - ancill related mods
_test.mf77 - radn related mods
rad_fix.mf77 - bug fix in 5.1 UM code
@]

We were unable to find these mods but have retrieved other 5.3 Moses2.2 mods from the Met Office system.

[@
amv0f503.mf77
amv0n503.mf90 - Bug fix for reconfiguration
amv0t503.mt

amv1f503.mf77 - Bug fixes for radiation code
amv1t503.mt

ang0h503.mh
ang0t503.mt

ang0n503.mf90 - Contains amv0n503.mf90
ang0t503.mt

ang0f503.mf77
ang0t503.mt
@]


!!Implementation options


We have the following options:

# '''Amend the Moses coastal tiling mods to work with the Moses version FAMOUS currently uses'''.
** Scientific concerns with older Moses. 

# '''Alter the original coastal tiling mods to work at 4.5 with the Moses2.2 mods from Martin Best.'''
** The main difficulty in converting from new to old dynamics is that the array dimensions have changed.
** Would involve altering the code by hand, line by line, which would be time-consuming exercise and may introduce errors.

# '''Write a wrapper to the 6.1 Moses2.2 code which already has coastal tiling included, so we could it plug into 4.5.'''
** This would require careful thought.
** Although the array dimensions have changed, the data is essentially in the same place.
** Other things to consider would be halos and the use of include files v comdecks.
** It may be advantageous for FAMOUS to use the newest version of Moses2.2.
** Ros has a script to convert code from 4.5 to 6.1 (include files and *CALLs etc) which could provide a good starting point.
to:
delete
Page last modified on February 02, 2007, at 02:23 PM by Annette