The NASA Modeling, Analysis and Prediction (MAP) Program

1
The NASA Modeling, Analysis and Prediction (MAP)
Program

• Don Anderson
• NASA Headquarters
• OCRT Meeting
• Goat Island
• April 11, 2006

2
2011 Overarching Vision

• a new approach is needed for climate model
  development. It is not practical for a single
  group to have expertise in all the components of
  a global model, as the models are becoming more
  complex and include more physical and biological
  processes. Models for long-term climate must
  extend from the deep ocean to the upper
  atmosphere, and components of the global system
  that were once treated as specified boundary
  conditions or neglected altogether must now be
  simulated and allowed to interact with other
  parts of the model.
• GSFC Earth-Sun Exploration Division 2006
  Strategic Plan

3
2011 Vision - Science Drivers

• Coupled Earth System models and advanced data
  assimilation systems
• System components fully interactive coupled
  ocean, atmosphere, land, sea-ice, chemistry,
  biosphere executing multiple what if scenarios
• Models easily reconfigured for short-term
  prediction (weather extreme events) to seasonal
  to long-term climate, global and regional
• Contributions to climate model development from
  diverse expert groups multiple discipline
  (chemistry, pollution transport, land, sea-ice,
  atmosphere) collaboration
• Implementation of 4D Variational data
  assimilation system with new data types
• Open standards
• Increased model analysis fidelity
• Enhanced resolution (horizontal, vertical,
  temporal, spectral)
• Estimates of Uncertainty/reliability
• large ensembles (perturbed initial conditions,
  multi-Center initial states, multi-Center models)
• Re-emphasis of data (satellite and in-situ
  observations for validation and simulation
  preparation for new missions)
• Greatly enhanced validation efforts with rapid
  access to data and model output from multiple
  Centers
• Observing System Simulation Experiments (OSSEs)
• Enhanced tools for testing new methods at reduced
  effort and cost
• Combinatorial problem in test and evaluation
  (many components, many versions requiring
  system-level validation)

4
MAP Modeling Environment Development Path
The Holy Grail
Fully Interactive Earth System Model
Multi-component Models
Component Models
Process Models
Knights of the NASA Earth Science Community
5
MAP NRA the MAP Modeling Environment(ME)
MAP Modeling Environment
Crosscutting Themes Focus Areas Model, Analysis,
Prediction Program / Multi-investigator proposals
ESMF
Core Integration Team

External NRA Proposals
GISS Model E
CMAI
GMI
ECCO II
6
MAP grant example Schematic of the Carbon Data
Assimilation SystemSome Element partners JPL,
Harvard, MIT, GISS, CSI, WHOI, Ames, MSFC DOE,
NSF, NOAA,NASAgt ESMF
7
GEOS-5 GCM STRUCTURE
HISTORY
CAP
COUPLED
ODAS
Data AGCM
Data OGCM
AGCM
OGCM
CICE
Poseidon
MOM4
MITogcm
Poseidon
OBIO
8
ESS Planned Hurricane Field Campaign
NAMMA-06 Out of Africa African Easterly
Waves are the progenitors of many late-season,
strong category hurricanes that strike the U.S.
Leverages off AMMA SOP-3 and will be based in
Cape Verde Multidisciplinary approach involving
Weather, Water Energy, Composition foci
Partnering with European consortium, NOAA HRD
NAMMA-06 science in line with CCSP objectives
Platforms to include DC-8, Aerosonde, TOGA NPOL
radars, micropulse lidars
African SAL Impact on tropical cyclogenesis -
hypothesize to be a hurricane suppressant, in an
ocean marginally suitable for hurricane
generation examine SAL impact on microphyiscs,
thermodynamic instability, shear,airmass
dessication
9
MAPME 2006 System Prototyping 1.
COMPUTING NODES
EXPERIMENT Embed WRF w/ GEOS5
DATA NODES
NASA Ames Mt. View, California
NAS MODEL OUTPUT PORTALS
MSFC OUTPUT PORTALS
SPoRT Facility NASA Marshall Huntsville, Alabama
All model Outputs
SIVO/NCCS Analysis Visualization
Prepare Map Movies Web Access
Ocean Color Group
WRF EXTERNAL MODEL OUTPUT PORTALS
Satellite Data
Graphics, Movies, Satellite Data
Other Model Output
SCIENCE ANALYSIS
10
MAPME 2006 System Prototyping 2.
DISTRIBUTED COMPUTING NODES
EXPERIMENT Configure GEOS5 w/ Different Ocean
Models
NASA Ames Mt. View, California
INTEL Corp Portland, Oregon
Northrop Grumman McLean, Virginia
NASA Goddard Greenbelt, Maryland
NEXT-GENERATION NETWORKS
Conventional Network (600 Mb/s)
National Lambda Rail (10 - 40 Gb/s)
NCCS/NAS MODEL OUTPUT PORTALS
DAAC DATA PORTALS
SCIENCE ANALYSIS
EXTERNAL COLLABORATORS
GFDL/NCDC MODEL OUTPUT PORTALS
Model to model model to data validation /
comparisons
11
Summary of Capabilities
ESMF Re-configurable models turn
on/off components, on/off outputs user settable
via web service
Cross-site workflow (data/jobs) Remotely
initiated via web schedule, monitor, manage with
greater autonomy across multi-site cyber-infrastru
cture
Development Environment
Experimentation Environment
Configure model, set up work flow all remotely
initiated via Web - ultimately SOA
Grey denotes Option turned off
Visualization Analysis Servers providing
on-demand support
Substantial compute resources connected
by high-speed networks load-balanced scheduled
vs. queued transparent site selection
Robust and non-intrusive infrastructure
support (security working across domains,
network performance QOS,)
12
Coupled Full-Earth Modeling and Data
Assimilation System for Earth-Sun Science
Integrating Earth System Modeling and Observations
2006 NASA MAP Modeling Environment
2005 Project Hurricane
GEOS-5
fvGCM (GEOS4)
JCSDA/GSI
Project Columbia
13
Dons Vision