evolution of modeling and data assimilation at nasa/gsfc

62
Evolution of Modeling and Data Assimilation at NASA/GSFC Arlindo da Silva Global Modeling and Assimilation Office, NASA/GSFC [email protected] CPTEC Workshop Cachoeira Paulista, Sao Paulo, Brazil 8-10 December 2008

Upload: oral

Post on 19-Mar-2016

53 views

Category:

Documents


1 download

DESCRIPTION

Evolution of Modeling and Data Assimilation at NASA/GSFC. Arlindo da Silva Global Modeling and Assimilation Office, NASA/GSFC [email protected]. CPTEC Workshop Cachoeira Paulista, Sao Paulo, Brazil 8-10 December 2008. Outline. - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Evolution of Modeling and Data Assimilation at NASA/GSFC

Evolution of Modeling and Data Assimilation at NASA/GSFC

Arlindo da SilvaGlobal Modeling and Assimilation Office, NASA/[email protected]

CPTEC WorkshopCachoeira Paulista, Sao Paulo, Brazil

8-10 December 2008

Page 2: Evolution of Modeling and Data Assimilation at NASA/GSFC

Outline GEOS-5 Earth Modeling System for

Prediction and Data Assimilation Scope and architecture: system of systems Historical overview and Roadmap Quick overview of sub-systems

The case for a component based, open system Rationale for frameworks: the ESMF Other frameworks: WRF, PRISM

Concluding Remarks

Page 3: Evolution of Modeling and Data Assimilation at NASA/GSFC

GEOS-5 Modeling Systems in support of NASA missions

Page 4: Evolution of Modeling and Data Assimilation at NASA/GSFC

4

Office HeadMichele Rienecker

Strategic Management TeamMax Suarez, Ron Gelaro, Steven Pawson,

Siegfried Schubert, Man-Li Wu, Arlindo da Silva, Gi-Kong Kim

AtmosphericAssimilation

Ron Gelaro

SubSeasonal-DecadalVariability & Prediction

Siegfried Schubert

ModelingMax Suarez

ConstituentsChem: Steven PawsonAero: Arlindo da Silva

Operational ProductsGi-Kong KIm

Civil Service Staff: 17 Contractor Staff: 52University Research Staff: 21

The Global Modeling and Assimilation Office (GMAO) is a component of the Earth Sciences Division at NASA's Goddard Space Flight Center. We contribute to NASA's Science Mission Directorate in the development and use of satellite observations through the integrating tools of models and assimilation systems.

Page 5: Evolution of Modeling and Data Assimilation at NASA/GSFC

• Contribute to Instrument Team products; advance the use of NASA data

• AURA: MLS, HIRDLS, TES, OMI• AQUA: MODIS, AIRS• CERES, CALIPSO• Field Campaigns (INTEX, NAMMA, TC4, ARCTAS, …)

• Science areas: • MERRA: reanalysis -hydrological cycle - NASA data in climate context• Prediction: weather, short-term climate, drought

• Aerosol-weather connections• Weather-climate connections• Chemistry-climate interactions

•Technical areas: • ESMF: improving extensibility of models through advanced software• GEOS-5 model: supports NASA’s MAP community

•Future missions: • OCO, SMAP, Aquarius• NPP - Joint Center for Satellite Data Assimilation• Decadal Survey - Wind Lidar mission (with Code 613.1)

Global Modeling & Assimilation OfficeGlobal Modeling & Assimilation OfficeContributing to NASAContributing to NASA’’s Missions Mission

Page 6: Evolution of Modeling and Data Assimilation at NASA/GSFC

Atmosphere Meteorological analyses (u,v,T, q): weather prediction, climate analyses Chemistry constituents: ozone, coupled with meteorology Chemistry constituents: CO, CO2 under development Aerosols: Transport, with source distributions from satellite GEOS-5 AGCM, currently 3Dvar, 4Dvar prototype in testing phase

Land Surface Soil moisture, surface temperature and snow Catchment LSM with EnKF

Ocean Retrospective Ocean analyses (u, v, T, S) for seasonal forecasts MOM4: OI, Assimilation in the CGCM coupled to atmospheric analysis Poseidon: EnKF Ocean color analyses: ocean time series, removing cross-satellite biases Poseidon: SEIK filter

Goal: Integrated Earth System Analysis, with consistent analyses across all components

GMAO Assimilation System(s)

Page 7: Evolution of Modeling and Data Assimilation at NASA/GSFC

04/24/23 7

MO

DE

LO

BSER

VATI

ON

SAN

ALYS

IS

4455 2 2 2.52.5 2 2 2.2.55 1 1 11 0.5 0.5 2/3 2/3 0.25 x 1/30.25 x 1/3

Aries Dynamical CoreAries Dynamical Core

NESDIS – Retrieved TOVS NESDIS – Retrieved TOVS TemperatureTemperature

TOVS/AMSU/AIRS RadianceTOVS/AMSU/AIRS Radiance

ScatterometerScatterometer

Conventional Observations (radiosondes, aircraft, …)Conventional Observations (radiosondes, aircraft, …)

Total Precipitable Total Precipitable WaterWater

OptimalOptimalInterpolationInterpolation Physical-Space Statistical Analysis SystemPhysical-Space Statistical Analysis System

GEOS-3GEOS-2 GEOS-4(FVDAS)

GEOS-1 GEOS-5

SSI SSI (NCEP)(NCEP)

1 1 1.21.255

GSI GSI (NCEP/NASA)(NCEP/NASA)

““GSFC” physicsGSFC” physics

Finite Volume CoreFinite Volume Core

NCAR NCAR physicsphysics

hybrid hybrid physicsphysics

MODIS windsMODIS winds

Total Precipitable Total Precipitable WaterWater

Evolution to GEOS-5 operational assimilation system

Page 8: Evolution of Modeling and Data Assimilation at NASA/GSFC

GEOS-5Atmosphere

GEOS-62011

GEOS-5AO system

• Coupled to LSM• ADAS + Adjoint tools• Replay for “coupling”• O3 assimilation• Coupled to GOCART• Coupled to GMI Combo

• Non-hydrostatic capable

• Physics for hi-res• Chem assim• ADAS: 4D-Var weak

constraint

Weather - Climate coupling

Chemistry-Climate coupling

Chem-weather prediction

ESMIESA

• Ocean• ODAS• LDAS

Short-term climate predictions

“Coupled” A-O analysesScience

IESA with the ocean

GEOS-5 Roadmap

Products - instr. teamsField campaigns, NWP

MERRAScience

Beginning of IESA

GEOS-5n(2008-10)

• AGCM w. hydrostatic cubed sphere

• ADAS: 4D-Var prototype• LSM with Dyn. Veg.• Carbon species assim.• Ocean biogeochem• Sea-ice

Climate changeMore Science

PIESA (weakly coupled, consistent analyses)

Page 9: Evolution of Modeling and Data Assimilation at NASA/GSFC

Brief overview of GEOS-5

Page 10: Evolution of Modeling and Data Assimilation at NASA/GSFC

10

AGCM Finite-volume dynamical core Bacmeister moist physics Physics integrated under the Earth

System Modeling Framework (ESMF) Generalized vertical coord to 0.01 hPa Catchment land surface model Prescribed aerosols Interactive ozone Prescribed SST, sea-ice

Analysis Grid Point Statistical Interpolation (GSI from NCEP) Direct assimilation of satellite radiance data using JCSDA Community Radiative Transfer Model (CRTM) Variational bias correction for radiances

Assimilation Apply Incremental Analysis Increments (IAU) to reduce shock of data insertion IAU gradually forces the model integration throughout the 6 hour analysis period

GEOS-5 Atmospheric Data Assimilation System

qn

t

total

dynamics(adiabatic) physics(diabatic) q

Model predicted change Correction from DASTotal “observed change”

Analysis

Background (model forecast)Raw analysis (from GSI)

Assimilated analysis(Application of IAU)

03Z 06Z 09Z 12Z 18Z15Z 21Z 00Z 03Z

Initial States for CorrectorAnalysis Tendencies for CorrectorCorrector Segment (1- and 3-hrly products)

Page 11: Evolution of Modeling and Data Assimilation at NASA/GSFC

11

NCEP/EMC-GMAO Code Managementfor Atmospheric Data Assimilation

Time

GMAO

EMC

* * EMC, GMAO System change Repository change+ Repository Merger (new tag)

* * * * * * * *

* * * * * * *

Repository

1 3Accepted changes

2

GSI & CRTM supported

Process: similar to ECMWF & Météo-Francewho have annual code mergers

But, to promote collaboration and transitions, EMC and GMAO use same repository and mergers are more frequent (3 months)

Protocols1 – EMC, GMAO take (agreed-upon) merged

code from repository to begin work2 – EMC, GMAO incorporate developments into repository3 – Code mergers, repository changes and

timing are NCEP’s decision

+ +

3 months

Page 12: Evolution of Modeling and Data Assimilation at NASA/GSFC

12

DATA SOURCE/TYPE PERIOD DATA SU PPLIER Conventional Data Radiosondes 1970 - present NOAA/NCEP PIBAL winds 1970 - present NOAA/NCEP Wind profiles 1992/5/14 - present UCAR CDAS Conventional, ASDAR, and MDCRS aircraft reports 1970 - present NOAA/NCEP

Dropsondes 1970 - present NOAA/NCEP PAOB 1978 - present NCEP CDAS GMS, METEOSA T, cloud drift IR and visible winds

1977 Š present NOAA/NCEP

GOES cloud drift winds 1997 Š present NOAA/NCEP EOS/Terra/MODIS winds 2002/7/01 - present NOAA/NCEP EOS/Aqua/MODIS winds 2003/9/01 - present NOAA/NCEP Surface land observations 1970 - present NOAA/NCEP Surface ship and buoy observations

1977 - present NOAA/NCEP

SSM/I rain rate 1987/7 - present NASA/GSFC/DAAC SSM/I V6 wind speed 1987/7 - present RSS TMI rain rate 1997/12 - present NASA/GSFC/DAAC QuikSCAT surface winds 1999/7 - present JPL ERS-1 surface winds 1991/8/5 Š 1996/5/21 CERSAT ERS-2 surface winds 1996/3/19 Š 2001/1/17 CERSAT

GEOS-5 input data streams

Page 13: Evolution of Modeling and Data Assimilation at NASA/GSFC

MERRAModern Era Retrospective-analysis for

Research and Applications supports NASA's Earth Science interests by1. Utilizing the NASA global data assimilation system to

produce a long-term (1979-present) synthesis that places the current suite of research satellite observations in a climate data context.

2. Providing the science and applications communities with state-of-the-art global analyses, with emphasis on improved estimates of the hydrological cycle on a broad range of weather and climate time scales

Page 14: Evolution of Modeling and Data Assimilation at NASA/GSFC

14

http://gmao.gsfc.nasa.gov/merra/Michael Bosilovich, Siegfried Schubert & Gi-Kong Kim

MERRA System

1/2° 2/3° 72L to .01 mb1979-presentGSI Analysis with IAUParallel AMIP run

EMPHASIS ON WATER CYCLE Global Precipitation, Evaporation, Land Hydrology, Cloud parameters and TPW

GLOBAL HEAT AND WATER BUDGETS FOR ALL PROCESSES

DIURNAL CYCLE FROM HOURLY 2-D FIELDS

Consistent 1979-present 3D aerosol time series will also be produced

MERRA

Page 15: Evolution of Modeling and Data Assimilation at NASA/GSFC

15

Precipitation (mm/day)January 2004 July 2004GEOS-5 GEOS-5

GPCPGPCP

Page 16: Evolution of Modeling and Data Assimilation at NASA/GSFC

16

Precipitation - GPCP (mm/day): July 2004

Page 17: Evolution of Modeling and Data Assimilation at NASA/GSFC

17

The adjoint (transpose) of a data assimilation system allows accurate and efficient estimation of observation impact on analyses and forecasts

determined with respect to observational data, background fields or assimilation parameters, all computed simultaneously

Adjoint Tools for Observation Impact StudiesRon Gelaro and Yanqiu

Zhu

impacts of arbitrary subsets of observations (e.g., separate satellites, channels or locations) can be easily quantified

GSI Analysis System

invisible

Forecast Model invisible

Input: Observations and

Background

Analyzed State

Output: Forecast

Adjoint GSI Analysis

System

Adjoint Forecast

ModelOutput:

Observation and Background Sensitivity

Analyzed State

SensitivityInput:

Forecast

Observation Impact

Page 18: Evolution of Modeling and Data Assimilation at NASA/GSFC

18

Evaluating AIRS Impact in GEOS-5

AIRS brings slightly positive impact on forecast skill in Northern Hemisphere; clear positive impact in Southern Hemisphere. Currently, forecast skills are increased when moisture channels from AIRS are not included…

Data from most AIRS channels improve the GEOS-5 forecast

Some AIRS channels degrade the forecast

Forecast Skill vs. Time

Control + AIRSControl

NH

SH

NH

Chan

nel I

ndex

24-hr Forecast Error Reduction vs. Channel

ControlControl + AIRS without moisture channels

Traditional Data Impact Studies Emerging Adjoint-based Tools

improve

degrade

Page 19: Evolution of Modeling and Data Assimilation at NASA/GSFC

1921 UTC 00 UTC 03 UTC

6-hour assimilation window

ooo

oo

oo

oo o

AnalysisObservations

Atmospheric Model/GCM Finite-volume dynamic core Bacmeister moist physics Physics integrated via ESMF Catchment land surface model Prescribed aerosols Interactive ozone

Atmospheric Analysis System Gridpoint Statistical Interpolation (GSI) TLM/Adjoint finite-volume dynamical core Direct assimilation of satellite radiances JCSDA Community Radiative Transfer Model (CRTM) Variational bias correction for radiances

GEOS-5 4D-Var Atmospheric Data Assimilation System

Page 20: Evolution of Modeling and Data Assimilation at NASA/GSFC

20

4D-Var Preliminary ResultsSingle Observation Experiments

t 0

Observation at the end of the 6-hr assimilation window

t -3h

t +3h

Page 21: Evolution of Modeling and Data Assimilation at NASA/GSFC

21

1. Use the hydrological catchment as the fundamental land surface unit.

Don’t assume land surface element has a shape defined by the overlying atmospheric grid

2. Within each catchment, use hydrologicalmodels for dealing with subgrid-scale soilmoisture distributions.

TOPMODEL, with a special treatment of the unsaturated zone. (We employ many of the ideas introduced by Famiglietti and Wood, 1994.)

GEOS-5: The NASA Catchment LSMGEOS-5: The NASA Catchment LSM

Page 22: Evolution of Modeling and Data Assimilation at NASA/GSFC

22

CANOPY RADIATIVETRANSFER

LAI & clumping profilesleaf albedo

PAR profiles, sunlit/shadednet SW to soil

patch albedo (canopy, soil, snow)

CANOPY BIOPHYSICSCi

Chl/N profilephotosynthesis=

Acan(leaf Chl, Ci, PAR, LAI,Tcan)conductance=

gcan(moisture,Tcan,height,VPD, Acan)

ALLOCATION/PHENOLOGY

budburst(Tgdd), cold/dry decidupdate individ C&N pools

plant respirationN uptake, N fixation

ALLOMETRY/GROWTH/REPRODupdate plant geometryestablish new seedlings

density dependencemortality

DISTURBANCEfire(above-ground biomass,

dryness(soil moisture)) combustion productslitter, new patches

PAR[layer]sunlit/shaded

Albedo, SW, CO2fire aerosolsVOCs

GCM ATMOSPHEREclimate

chemistry

Sensible & latent heatmomentum

P, VP, CO2Tair, PrecipSW , PARbeam/diffuse

u,v, P, VPTair , LWPrecip

DGTEM

SOIL BGClabile C, labile N

available Nslow C, slow N

soil respiration= (substrate, moisture, Tsoil)

LANDSCAPE & VEG STRUCTUREpatch (age distrib) cohort (density) individual plant functional type (pft) plant mass C&N:foliage, stem, root C&N: labile storage plant geometry LAI, SLA profile, dbh, height, root depth crown size (axes)

litter

N

net CO2 uptake [layer]

updatestructure

hourlyseasonal-decadal

ENT Dynamic Global Terrestrial Ecosystem

Model(Kiang, Koster, Moorcroft, Ni-Meister,

Rind)

SVAT: LAND SURFACE ENERGY & WATER BALANCEcanopy energy balance

soil energy balancesoil moisture

snow cover, snow albedosoil albedo

conductancenet SW

landscapeand veg structure

Tsoil, Tcanopy snow albedosoil albedo, soil moisture

mixed canopies

coupled C&N

daily carbon

ED

deep soillayer

Page 23: Evolution of Modeling and Data Assimilation at NASA/GSFC

23

Assimilation product agrees better with ground data than satellite or model alone.Modest increase may be close to maximum possible with imperfect in situ data.

Reichle et al., JGR, 2007Anomaly time series correlation coeff. with in situ data [-] (with 95% confidence interval)

Confidence levels: Improvement of assimilation over

N Satellite Model Assim. Satellite Model

Surface soil moisture 23 .38±.02 .43±.02 .50±.02 >99.99% >99.99%

Root zone soil moisture 22 n/a .40±.02 .46±.02 n/a >99.99%

Global assimilation of AMSR-E soil moisture retrievalsGlobal assimilation of AMSR-E soil moisture retrievals

Validate with USDA SCAN stations(only 23 of 103 suitable for validation)

Soil moisture [m3/m3]

Assimilate retrievals of surface soil moisture from AMSR-E (2002-06) into NASA Catchment model (GEOS-5)

Page 24: Evolution of Modeling and Data Assimilation at NASA/GSFC

24

RadiativeModel

(OASIM)

CirculationModel

(PoseidonV2)

BiogeochemicalProcesses Model

Winds SST

Layer DepthsIOP

Ed(λ)Es(λ)

Sea Ice

NASA Ocean Biogeochemical Model (NOBM)Winds, ozone, relative humidity, pressure, precip. water, clouds (cover, τc), aerosols (τa, ωa, asym)

Dust (Fe)

Advection-diffusion

Temperature, Layer Depths

Ed(λ) Es(λ)

Chlorophyll, Phytoplankton GroupsPrimary ProductionNutrientsDOC, DIC, pCO2

Spectral Irradiance/Radiance

Outputs:

Page 25: Evolution of Modeling and Data Assimilation at NASA/GSFC

25

Biogeochemical Processes Model

Diatoms

Chloro-phytes

Cyano-bacteria

Cocco-lithophores

Si

NO3

NH4

Herbivores

N/CDetritus

Fe

SilicaDetritus

PhytoplanktonNutrients

IronDetritus

Ecosystem Component

N/CDetritus

Phyto-plankton

DissolvedOrganicCarbon

Dissolved InorganicCarbon

pCO2(water)

pCO2(air)

Winds,Surface pressure

Carbon Component

Herbivores

Page 26: Evolution of Modeling and Data Assimilation at NASA/GSFC

26Lars Nerger, “Assimilation of SeaWiFS Ocean Chlorophyll data with a simplified SEIK filter”

Page 27: Evolution of Modeling and Data Assimilation at NASA/GSFC

Chemistry and Aerosols The current GEOS-5 aerosol/chemistry

capabilities evolved from several off-line CTM efforts at/through Code 613.3: GOCART aerosols, CO/CO2 (Chin et al.) CARMA aerosol microphysics (Toon et al.,

through Colarco) StratChem (Douglas, Stolarski et al.) GMI Tropospheric+Stratospheric (Combo)

Chemistry Which in turn derives from Harvard GEOS-Chem

and StratChem

Page 28: Evolution of Modeling and Data Assimilation at NASA/GSFC

Aerosol Modeling at GMAOAerosols transported on-line within GMAO’s

Climate/Forecasting models In climate mode: no data assimilation In replay mode, using assimilated meteorology

Aerosols transported on-line within the GCM, without need for time interpolation of winds/diagnostics

Can be used for aerosol data assimilation In full assimilation mode, combined

meteorological/aerosol assimilationEffective way of dealing with contamination of

TOVS/AIRS radiances by aerosols

Page 29: Evolution of Modeling and Data Assimilation at NASA/GSFC

Aerosol Processes by GEOS-5 Advection:

Same Lin-Rood used my many off-line CTMs Diffusion:

GEOS-5 has Lock type PBL parameterization Convective transport:

Relaxed Arakawa-Schubert (RAS) parameterization RAS provides convective transport as well as scavenging

Aerosol direct effects: Chou et al. radiation package Model transports dry aerosol mass; RH hygroscopic growth

included during Mie calculation Indirect effects (not yet integrated):

Nenes and Seinfeld parameterization for water clouds; additional ice clouds paramerization(Y.Sud)

Page 30: Evolution of Modeling and Data Assimilation at NASA/GSFC

Collaborator: Mian Chin, Code 613.3

Page 31: Evolution of Modeling and Data Assimilation at NASA/GSFC

GEOS-5/GOCART Forecasts

Global 5-day chemical forecasts customized for each campaign O3, Aerosols, CO, CO2,.. Tag tracers

Driven by real-time biomass emissions from MODIS

Pre-mission System customization

During-mission Web visualization, data delivery In-field forecasting support Comparison to aircraft data

Post-mission: Gridded datasets available

online for post mission analysis In depth evaluation, model

tuning A truly GSFC wide effort:

GMAO, ACDB, SIVO, NCCS

CO

Smoke

SO4

O3

Page 32: Evolution of Modeling and Data Assimilation at NASA/GSFC

Dust + Sea Salt

GEOS-5/GOCART

ECMWF

Forecast Valid at3Z 20 July 2008

Page 33: Evolution of Modeling and Data Assimilation at NASA/GSFC

Aerosol Data Assimilation at GMAOEmphasis on estimation of

Global, 3D aerosol concentrationsAerosol sources and model parametersObserving System Simulation

Experiments (OSSE)Aerosol effects on climate, focus on

hydrologic cycleAerosol forecasting capability in support

of field campaigns

Page 34: Evolution of Modeling and Data Assimilation at NASA/GSFC

MODIS Radiances 1D-Var scheme using GOCART aerosol

fields as background (Weaver et al 2005) Ocean: draws to all 7 MODIS channels,

drawing the tighest to 870nm Land: draws only to 466 nm

Algorithm not integrated into GMAO’s realtime aerosol forecasting system

Page 35: Evolution of Modeling and Data Assimilation at NASA/GSFC
Page 36: Evolution of Modeling and Data Assimilation at NASA/GSFC

OMI radiancesNext step: extension of 1D-Var

scheme will for assimilation of OMI radiancesCombined assimilation of MODIS/OMI radiancesBuilt in adaptive bias correction for homo- genizing observing system

Page 37: Evolution of Modeling and Data Assimilation at NASA/GSFC

CALIPSO

Simulation of attenuated backscatter from 3D aerosol distributions

CALIPSO aerosol 1D-Var at model vertical resolution

Adaptive tuning of GEOS-5 PBL Joint assimilation of MODIS/OMI/CALIPSO

measurements

Page 38: Evolution of Modeling and Data Assimilation at NASA/GSFC

Biomass Emissions Near real time estimates based on MODIS Fire

products (AQUA/TERRA) Used extensively during field campaigns

Currently developing next generation algorithm: Based on fire radiative power Determination of flaming/smoldering ratios and fire areas Injection layer determined by Saulo’s Plume

Rise parameterization Focus on MODIS, eventually geostationary Starting collaboration with NOAA and NRL

Page 39: Evolution of Modeling and Data Assimilation at NASA/GSFC

39

Ozone in GEOS-5 DAS

Data: – SBUV and OMI ozone– TOVS and AIRS radiances– MLS retrieved stratospheric ozone profiles

Model:– Parameterized chemistry (production and loss

rates)

Prognostic ozone used in:– Radiative heating computations in AGCM– Assimilation of IR radiances

Page 40: Evolution of Modeling and Data Assimilation at NASA/GSFC

NOAA 16 SBUVMLS

SBUV daytime only – no data near South Pole due to high solar zenith angleMLS orbital limit ±82º

Assimilating AURA/MLS ozone

Ozone hole develops in MLS

assimilation

Ozone partial pressure (mPa)

Zonal mean ozone 9/30/2004 00UTC

MLS only

Meta Sienkiewicz and Ivanka Stajner

SBUV/2 only

Page 41: Evolution of Modeling and Data Assimilation at NASA/GSFC

41

High-resolution chemistry-climate model simulation with GEOS-5

Coherent filaments are peeled from the edge and interior of the polar vortex

GEOS-5 AGCM with Stratospheric chemistry module from GSFC/ACD• Simulations at 0.666° 0.5° with 72 layers • Year is defined only by boundary conditions (SST, ice, chemical emissions) • Example: April 1, “2004” - 70hPa near end of cold simulated Arctic winter

Page 42: Evolution of Modeling and Data Assimilation at NASA/GSFC

42

Research Goals: • Improved representation of processes• Model evaluation in context of carbon-cycle • Model-data combinations • Combining assimilation and inversion• Forward modeling and sampling for OSSEs (preparing for OCO)

Programmatic Goals: • GEOS-5: tool for carbon-cycle science • Earth-system modeling and assimilation

Products:• Assimilated CO2 concentrations based on existing (EOS) satellite data

• 2002 onwards: based on AIRS, MODIS, …• 2008 onwards: based on OCO plus others

• Flux estimates derived using inversion methods• Spatio-temporal resolution determined in project

• Extensions to GEOS-5 for carbon-cycle science • GCM couplings to land and ocean • Extended assimilation capabilities

Carbon Data Assimilation - A Brief OverviewSteven Pawson

GSFC, CSU, ORNL, WHOI, HU

Page 43: Evolution of Modeling and Data Assimilation at NASA/GSFC

Software Development and External Collaborations

Page 44: Evolution of Modeling and Data Assimilation at NASA/GSFC

Modern models integrate components from different sources ESMF accelerates development cycle

NASA AGCM for climate and weather

LANL sea ice model

Add in the assimilation components and the satellite data science + future mission design

GMAO ocean biology

NSF NCAR / NASA GSFC / DOE LANL ANL / NOAA NCEP GFDL / MIT / U MICH

GMI chemistry

GOCART aerosol

GMAO Physics

GFDL Dynamics

GFDL Ocean GMAO Land

GMAO Ocean Biology

Page 45: Evolution of Modeling and Data Assimilation at NASA/GSFC

GEOS-5 Component Architecture

Page 46: Evolution of Modeling and Data Assimilation at NASA/GSFC

GEOS-5 AGCM at a GlanceAGCM

Dynamics Column Processes

Moist Physics

SurfaceProcesses

Turbulence

Radiation

AerosolChemistry

Dynamical Core

Gravity Wave Drag

The Aerosol/Chemistry componentmust provide the following toradiation: Ox, O3, CH4, N2O, CFC11, CFC12,CFC22 and aerosols (dust, seaSalt, SO4 and carbonaceous)

Page 47: Evolution of Modeling and Data Assimilation at NASA/GSFC

The AeroChem ComponentAerosol/

Chemistry

GOCARTAerosols

GSFC StratChemistry P & L GMI Combo

Strat/Tropo

Dust

Sea Salt

Sulfates

Carbonaceous

CO & CO2

Ox, O3

CH4

N2O

CFC11, CFC12, CFC22

etc.

Ox, O3

CH4

N2O

CFCs

Aerosols (Michigan)

etc

Ox, O3

CH4

N2O

CFCs

Data Aerosols

Age of Air

At runtime one selects one or more packages to run, and in case ofambiguity,which package provides a specific input to radiation

Page 48: Evolution of Modeling and Data Assimilation at NASA/GSFC

System of systems From such a collection of components several

systems can be developed Atmospheric data assimilation system Oceanic data assimilation system Seasonal forecasting system Coupled climate-chemistry system

The development of each subsystem require careful validation for the applications at hand.

Computational resources dictate the particular combination of complexity/resolution that can be exercised.

Page 49: Evolution of Modeling and Data Assimilation at NASA/GSFC

A High-Performance Framework for Earth Science Modeling & Data Assimilation

Pilot Project: 2002-2005Principal Investigators:

Core ESMF: Tim Killeen (NCAR) Modeling: John Marshall (MIT)Data Assimilation: Arlindo da Silva (NASA)

NASA/GSFC

Page 50: Evolution of Modeling and Data Assimilation at NASA/GSFC

Technological Trends In climate research and NWP...

increased emphasis on detailed representation of individual physical processes; requires many teams of specialists to contribute components to an overall coupled system

In computing technology...

increase in hardware and software complexity in high-performance computing, as we shift toward the use of scalable computing architectures

Page 51: Evolution of Modeling and Data Assimilation at NASA/GSFC

Community Response Modernization of modeling software

Abstraction of underlying hardware to provide uniform programming model across vector, uniprocessor and scalable architectures

Distributed development model characterized by many contributing authors; use of high-level language features for abstraction to facilitate development process

Modular design for interchangeable dynamical cores and physical parameterizations, development of community-wide standards for components

Development of prototype frameworksGFDL (FMS), NASA/GSFC (GEMS), NCAR/NCEP (WRF), NCAR/DOE (MCT), etc.

The ESMF aims to unify and extend these efforts

Page 52: Evolution of Modeling and Data Assimilation at NASA/GSFC

Objectives of the ESMF1. Facilitate the exchange of scientific codes

(interoperability)2. Promote the reuse of standardized technical

software while preserving computational efficiency3. Focus community resources to deal with changes

in computer architecture4. Present the computer industry and computer

scientists with a unified and well defined task5. Share overhead costs of the housekeeping aspects

of software development6. Provide greater institutional continuity to model

development efforts

Page 53: Evolution of Modeling and Data Assimilation at NASA/GSFC

Scientific BenefitsESMF accelerates

advances in Earth System Science1. Eliminates software barriers to collaboration among

organizations Easy exchange of model components accelerates

progress in NWP and climate modeling Independently developed models and data

assimilation methods can be combined and tested Coupled model development becomes truly

distributed process Advances from smaller academic groups easily

adopted by large modeling centers

Page 54: Evolution of Modeling and Data Assimilation at NASA/GSFC

Scientific Benefits, cont.ESMF accelerates

advances in Earth System Science2. Facilitates development of new interdisciplinary

collaborations Simplifies extension of climate models to upper

atmosphere Accelerates inclusion of advanced

biogeochemical components into climate models Develops clear path for many other communities

to use, improve, and extend climate models Many new model components gain easy access to

power of data assimilation

Page 55: Evolution of Modeling and Data Assimilation at NASA/GSFC

Design PrinciplesModularity data-hiding, encapsulation, self-sufficiency;Portability adhere to official language standards, use

community-standard software packages, comply with internal standards

Performance minimize abstraction penalties of using a framework

Flexibility address a wide variety of climate issues by configuring particular models out of a wide choice of available components and modules

Extensibility design to anticipate and accommodate future needs

Community encourage users to contribute components, develop in open source environment

Page 56: Evolution of Modeling and Data Assimilation at NASA/GSFC

Earth System Modeling Framework (ESMF)

1. ESMF provides an environment for assembling geophysical components into applications.

2. ESMF provides a toolkit that components use to

i. increase interoperabilityii. improve performance

portabilityiii. abstract common services

ESMF InfrastructureData Classes: Bundle, Field, Grid, Array

Utility Classes: Clock, LogErr, DELayout, Machine

ESMF SuperstructureAppDriver

Component Classes: GridComp, CplComp, State

User Code

Page 57: Evolution of Modeling and Data Assimilation at NASA/GSFC

Organizations using ESMF NASA

GEOS-5: all components under ESMF NOAA

NCEP operational GFS are ESMF compliant Next unified global/regional system being developed

under ESMF GFDL: FMS can co-exist with the ESMF

US Navy ESMF being used to couple models from costal to

global scales NCAR: CCSM adopting ESMF

Page 58: Evolution of Modeling and Data Assimilation at NASA/GSFC
Page 59: Evolution of Modeling and Data Assimilation at NASA/GSFC

Other frameworks Weather Forecast Research (WRF)

Can be wrapped as a single ESMF component and coupled to other models

ESMF components not interoperable with WRF components

Flexible Modeling System (GFDL) Architecture similar to ECMWF High degree of interoperability with ESMF

PRISM European counterpart of he ESMF Emphasizes flux-coupler aspects, no similar

infrastructure

Page 60: Evolution of Modeling and Data Assimilation at NASA/GSFC

http://prism.enes.org 60

PRISM vs ESMF

coupling superstructure

infrastructure software

User code

Running environmentPRISM

ESMF

Earth System Model

Page 61: Evolution of Modeling and Data Assimilation at NASA/GSFC

ESMF GovernanceExecutive BoardStrategic DirectionOrganizational ChangesBoard Appointments

Interagency Working GroupStakeholder LiaisonProgrammatic Assessment & Feedback

Advisory BoardExternal Projects CoordinationGeneral Guidance & Evaluation

Functionality Change Requests

Joint Specification TeamRequirements DefinitionDesign and Code ReviewsExternal Code Contributions

Implementation Schedule

Resource Constraints

Collaborative DesignBeta Testing

Working Project

ExecutiveManagement

Reporting

Reporting

weekly

Core Development TeamProject ManagementSoftware DevelopmentTesting & MaintenanceDistribution & User Support

daily

annuall

Change Review BoardDevelopment PrioritiesRelease Review & Approval

quarterly

Page 62: Evolution of Modeling and Data Assimilation at NASA/GSFC

Concluding Remarks Increased emphasis on detailed representation

of individual physical processes and inclusion of many earth system components Climate/NWP problem is too large for a single organization Emphasis on modeling environments where several models can be

derived from shared components to address problem with very diverse time and spatial scales

The adoption of software frameworks such as the ESMF simplifies the technical aspects of exchanging earth system components Sustainable collaborations must continue beyond the

initial code hand-out