Experimental El NioSouthern Oscillation Predictions by the UCLA Atmospheric General Circulation Mode

1
Experimental El Niño/Southern Oscillation
Predictions by the UCLA Atmospheric General
Circulation Model (GCM) Coupled to the MIT and
POP Oceanic GCMs using the Earth System Model
Framework (ESMF)

• C. R. Mechoso (1), Gabriel Cazes-Boezio (1, 3),
• J. A. Spahr (1), D. Menemenlis (2), and S.J. Zhou
  (4)
• Dept. Atmospheric and Oceanic Sciences, UCLA, USA
• NASA/Caltech Jet Propulsion Laboratory, USA
• On leave from IMFIA, Universidad de la Republica,
  Uruguay
• Software Integration and Visualization Office,
  NASA/GSFC and Northrop Grumman

This talk presents the first independent
adoption of ESMF technology.
2
Atmospheric General Circulation Model (AGCM)
Atmospheric Chemical Tracer Model (ACTM)
Data Broker
Oceanic General Circulation Model (OGCM)
Oceanic Chemical Tracer Model (OCTM)
UCLA Earth System Model

• Model components
• UCLA Atmospheric GCM
• LANL Parallel Ocean Program (POP)
• UCLA ACTM (which can include up to 64 species)
• Simple NASA/JPL Ocean Chemical Transport Model
• Distributed Data Broker

3
Increasing the Interoperability of an Earth
System Model Atmospheric-Ocean Dynamics and
Tracer Transports NCC4-624C. R. Mechoso, PI

• Major thrusts of the project
• to further our understanding of an ability to
  predict the dynamic interaction of physical
  and chemical processes affecting Earth
• to incorporate the use of NASA data and highlight
  its importance
• to demonstrate interoperability of codes used in
  the community of Earth Science.

4
UCLA AGCM - MIT OGCM
5
UCLA AGCM - MIT OGCM
6
UCLA AGCM - LANL POP
7
UCLA AGCM - LANL POP
8
Atmospheric General Circulation Model (AGCM)
Atmospheric Chemical Tracer Model (ACTM)
Data Broker
ESMF
Oceanic General Circulation Model (OGCM)
Oceanic Chemical Tracer Model (OCTM)
UCLA Earth System Model
Model components UCLA AGCM with upgraded PBL
parameterization LANL Parallel Ocean Program
(POP) and MIT OGCM, both in a quasi-global domain
with the same grid. UCLA ACTM (which can
include up to 64 species) Simple NASA/JPL
Ocean Chemical Transport Model
9
Porting Strategy

• Model codes (AGCM, OGCM) were restructured to
  isolate Initialize, Run and Finalize tasks (i.e.,
  were made ESMF compliant)
• An ESM Driver Program (EDP) was created to
  control the sequence in which those
  Initialize-Run-Finalize tasks and data transfers
  are executed, as well as to keep track of
  simulation time.

10
Coupled Atmosphere-Ocean Application in ESMF-
Superstructure
ESM Application
Atmos to Ocean Coupler
Ocean to Atmos Coupler
AGCM App
OGCM App
Initialize Run Finalize
Initialize Run Finalize
MIT OGCM POP OGCM
UCLA AGCM
Import/Export States
Import/Export States
Infrastructure
11
The difference in prediction between with and
without ESMF is possibly due to the different
regridding methods --- to be verified!
12
Generic Coupler in ESMF Observation couplers
for UCLA AGCM/LANL OGCM and for UCLA AGCM/MIT
OGCM are almost identical
13
Issues in Coupling Model Components(e.g., From
Ocean to Atmosphere)
export
import
export
import
Atmo Init1_a() Init2_a() Run_a() Final_a()
ssta
Ocean Init1_o() Init2_o() Init3_o() Run_o()
Final_o()
Sst_o
ua
S_o
va
U_o
Coupler Regrid()
V_o
Sea_o
F_o

• Function names in model components are
  different
• ESMFs solution
• Uses three standard functions with a stage
  option, initialize, run, and finalize
• Use a function registration service,
  setEntryPoint and setService
• The names of import variables of a use
  component are different from those of export
  variables of a provide component.
• The number of import variables of a use
  component are different from that of export
  variables of a provide component.

14
Generic Coupler design(automatically name
mapping and regridding)
CF
name
pointer
Atmo
Ssta
Destination field
needed source field
T_surf
ssta
Ua
Ptr_Ssta
Atmos
ssta
Sst_o
U_surf
ua
Va
Ptr_ua
register
ua
U_o
V_surf
va
Ptr_va
va
V_o
Match name
T_surf
Sst_o
Ptr_Sst_o
S_surf
S_o
Ptr_S_o
remake export state
Output import state
U_surf
U_o
Ocean
Ptr_U_o
V_surf
V_o
Ptr_V_o
regrid
Ptr_Sst_o
Ptr_Ssta
Sea_level
Sea_o
Ptr_Sea_o
Ptr_U_o
Ptr_ua
frazil
F_o
Ptr_F_o
Ptr_V_o
Ptr_va
For the case of only interpolation operation is
needed.
15

• SUMMARY
• The coupled atmosphere-ocean model shows skill in
  ENSO prediction from six months in advance. The
  skill is higher for the UCLA AGCM/MIT OGCM
  combination.
• The UCLA AGCM was coupled to the MIT OGCM and
  LANL POP using ESMF services.
• The most time demanding task was to make the
  model components ESM compliant.
• Much more work with the ESMF is needed for
  debugging, increased capability and user
  friendliness.
• Model codes integrated into the ESMF require
  maintenance if the framework is to become a
  standard for Earth System Modeling.

16
Project Organization

• Tier I Upgrade the UCLA Earth System Model
  (ESM). The principal upgrades are in the
  planetary boundary layer (PBL) parameterization
  of the AGCM and domain extension of the OGCM.
  Perform ENSO predictions.
• Tier II Address issues of code
  interoperability by using the ESMF services to
  couple the UCLA AGCM with either LANL POP or the
  MIT OGCM and by performing forecasts of El
  Niño/Southern Oscillation (ENSO).
• Tier III Assess the impact of NASA data by
  comparing ENSO forecasts using initial states
  provided by JPLs ECCO project (http//ecco.jpl.na
  sa.gov). The MIT OGCM is a component in the
  ECCOs data assimilation system, while POP is not.

17
The AGCM as ESMF Compliant

• Module AGCM_GridComp
• AGCM_Initialize
• AGCM_Run(start time, end time)
• AGCM_Finalize

18
ESM Application
Register components Create data bases
Initialize components
Call gridded components drivers (time advance)
Call coupler components for regridding and
transfers
Call gridded components to perform finalize
functions
19
Component Driver in EDP (e. g., AGCM)
Create ESMF grid Create fields and attach them to
import/export states
Read control files Set up node distribution
Retrieve SST from import state
Advance in time
Strore updated fields in export state
Clean up and close communications
20
Modeling Focus PBL in UCLA AGCM

• Traditional Framework
• Suarez et al. (1983)
• Revised Framework
• Konor and Arakawa (2005)

21
Coupler
AGCM
OGCM
CPL Ctrl
Time execution
E
I
OGCM Exec
AGCM Exec
Data Fields
Data Fields
CPL Ctrl
E
I
I
Data Reference
Import State
Data Flow
E
Export State
22
Double ITCZ Problem Hypothesis 1 Poor heat
transport by ocean eddied from upwelling regions
- Insufficient OGCM resolution? Hypothesis 2
Poor simulation of the zonal circulation -
Difficulties in the simulation of resolved and
subgrid processes?
Annual Mean SST Model
Observation
UCLA AGCM - MIT OGCM
23
ESMF Class Structure
ApplicationComponent
GriddedComponent
CouplerComponent
Superstructure
State
Bundle
Regrid
Infrastructure
Field
Grid
PhysGrid
DistGrid
F90
Data
Communications
Layout
Array
Comm
Route
C
MachineModel
Utilities TimeMgr, Config, LogErr, I/O etc.
24
ESMF Component Registration

• Components provide a single externally visible
  entry point, which register the other entry
  points with the ESMF. Components can
• Register one or more Initialization, Run,
  Finalize, and Checkpoint entry points.
• Register a private data block which can contain
  all data associated with this instantiation of
  the Component.

Higher level Comp
ESMF Framework Services
cmp_final()
cmp_register()
cmp_run()
cmp_init()
Public subroutine Private subroutine
25
Coupler ComponentsImport/Export States
The AGCM does not have access to the internals of
the OGCM, and vice versa. The exchange data is
through a coupler component, which exchanges the
roles of Export State from one component to
Import State for the other component. The coupler
components also do the grid transformations.
coupler
ocn_component subroutine ocn_run(comp,
ImportState,ExportState, Clock, rc)
atm_component subroutine atm_run(comp,
ImportState,ExportState, Clock, rc)
26
(No Transcript)
27
(No Transcript)