Earth System Modeling at UCLA Applications and Further Developments

Carlos Roberto Mechoso

Department of Atmospheric Sciences, University of California Los Angeles, USA

Goal: Develop and apply to problems of climate change a model that describes the coupled global atmosphere - global ocean system, including chemical tracers. Each model component designed for high-performance parallel internal execution and as a Common Architecture Component (CCA), and integrated in NASA’s ESM Framework.

www.atmos.ucla.edu/~mechoso/esm

Model elements:

• UCLA AGCM with upgraded PBL parameterization

• LANL version of Parallel Ocean Program (POP)

upgraded to hybrid vertical coordinates (HYPOP)

• UCLA ACTM (which can include up to 64 species)

• JPL Ocean Chemical Transport Model

• MIT and MOM OGCMs as options

UCLA Earth System ModelStage 3

Atmospheric General Circulation

Model (AGCM)

Data Broker

Oceanic General

Circulation Model (OGCM)

Atmospheric Chemical Tracer

Model (ACTM)

Oceanic Chemical Tracer

Model (OCTM)

Stratocumulus clouds play keyroles in the climate system.

Global climate models have serious difficulties in simulating these clouds.

ENSO simulations and predictions can be greatly improved by improving model performance with tropical stratocumulus decks.

UCLA AGCM v7.5 R. Terra, 2002

Diurnal Cycle in Amazonia

In Amazonia during the monsoon season, rainfall peaks at around 2 pm and convection peaks a few hours later. 2:00pm

4:00pm

The model captures the early afternoon peak of precipitation over the continental areas with

monsoon circulations

The highly realistic ENSO produced by the coupled AGCM/OGCM was obtained through extensive model development and numerical experimentation allowed by the efficient model codes developed under this project. (Contour interval: 0.50C left, 0.250C right )

140 160 180 200 220 240 260 280Longitude

3938373635343332313029282726252423222120

-1.50 -1.25 -1.00 -0.75 -0.50 -0.25 0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00

Sim

ula

ted

Yea

r

Sea surface temperature at equator

Below average Above average

A particular event Events in 20 years

Characteristics/assumptions:Multiple layers near the surface are assigned for the PBL

Resolves low-level vertical wind shear associated with low-level baroclinicityIncludes effects of cumulus cloud roots

The PBL-depth is predicted (Suarez et al., 1983)Easier to maintain PBL top discontinuity

Within the PBL, a s-coordinate ( . ., 1983)is used Suarez et al- PBL top and Earth surface are coordinate surfaces

Properties of PBL quantities are nearly uniform along coordinate surfacesT he processes near the PBL top are easier to formulate

A bulk formulation ( . )for convective eddy effects Randall et al unpublishedA -local K closure formulation for diffusive eddy effects

Vertical profiles of PBL quantities can deviate from purely mixed profiles

E

A Multi-layer PBL parameterization(Konor and Arakawa, 2001)

Thirty-five year long simula-tions with the UCLA ESM yielded a comprehensive 3-D picture of the evolution in the atmosphere of 20 million tons of chlorofluorocarbons (CFCs) released since 1965, and subsequent chemical reactions.

By simulating the full stra-tosphere rather than just the lower stratosphere where most CFC break-up occurs, the results significantly narrow uncertain-ties in CFC lifetimes.

National Aeronautics and Space Administration

Earth Science Enterprise

The First Comprehensive Picture of Chlorofluorocarbons in the Atmosphere

Percent CFC-11 emissions rate for 1965–1996. The highest emissions cluster over major industrial areas in the eastern United States and western Europe. (M. Gupta, R. P. Turco, C. R. Mechoso, and J. A. Spahr, 2001, JGR - Atmospheres.)

CFC-12 Evolutions in Selected Locations

Validation for the simulations came from a network of eight surface observation stations, including the five Advanced Global Atmospheric Gases Experiment (AGAGE) stations.

The results typically differed by just a few percentage points from each station’s readings. After 1997, all simulations show decreases in CFC-12, while the data show continued increases.

Methodology for concurrent execution of ESM components

Atmospheric surface fluxes that drive the ocean are produced by the model component known as AGCM/Physics; sea surface temperatures (SSTs) that drive the atmosphere are produced by the ocean model component known as OGCM/Baroclinic.

Coupling between ESM components

• Data transfers between ESM components are handled by Data Brokers.

• One type is a Centralized Data Broker (CDB), which assembles the entire data field from a producer on a single processor before sending it to consumers.

• CDBs can introduce performance bottlenecks and have processor memory limitations.

• In view of these considerations, we have developed the UCB/UCLA Distributed Data Broker.

Distributed Data BrokerFunction

To handle distributed data exchanges between the ESM component.

Major Tasks

•Gather information from each model component (may have been decomposed into many subdomains running at different processors).•Convert data resolutions, units…, etc. and redistribute them to the needed model components.•Keep track of coupling sequence, such as how often AGCM needs sea surface temperature, and which processors have those information.

Distributed Data Broker (DDB)Components

• Communications LibraryGeneral communication routines to manage the data exchanges based on standard communication toolkits (PVM(current), MPI (proposed), etc).

• Model Communications Library Callable routines for registering and exchanging information between model components.

• Data Translation LibraryRoutines for data regridding.

Reported Performance: 40 GFLOPS (2001)

UCLA ESM Application Component

Coupler Component

UCLAAtmospheric

ModelComponent

CCSMCoupler

LANLOceanModel

Component

UCLAChemistry

and AerosolModel

LANLSea-IceModel

Component

UCLA ESM Component Structure

Check-Point

• Code optimization and parallelization completed to a level acceptable with (1) research objectives, (2) available computer resources.

• Case studies completed that demonstrate good model performance.

• Perceived need to “modularize” major model components in order to facilitate implementation of model upgrades designed both in house and at other development groups.

Earth System Modeling Application

NCAR/LANL CCSM

NCEP Forecast

ESMF_GridComp

NCEP Forecast

ESMF_GridComp

MIT OGCM ocean

ESMF_CplComp Coupler Component performs data transformations and transfers between atmosphere and ocean Components

UCLA AGCM atmosphere

ESMF_AppCompCoupled atmosphere / ocean simulation driver

CCA ESM Prototype

Palette

Process Set A

Process Set B

Arena

DDB

Ocean

Atmos

GO Run Ocean

Run Atmos

Driver

GO

Run Ocean

Run Atmos

Driver

Run Atmos Produce

Consume

Atmos

Produce

Consume

Run Ocean

Ocean

Produce

Consume

DDB

Produce

Consume

DDB

Driver

Ocean

Atmos

DDB

Driver

Summary• An ESM (ocean-atmosphere dynamics and

chemistry) has been developed for high performance computers.

• The ESM includes a novel Distributed Data Broker, a flexible, high-level syntax for coupling components of multidisciplinary applications.

• The ESM is been applied to ENSO predictability and the evolution of chemical tracers (CFCs, Rn) in the atmosphere.

• The ESM code is been integrated in NASA’s ESM Framework. Model components are been designed as Common Architecture Components (CCA).