Title: Aura
1National Aeronautics and Space Administration Jet
Propulsion Laboratory California Institute of
Technology
Aura Microwave Limb Sounder David T. Cuddy Jet
Propulsion Laboratory California Institute of
Technology
October 27, 2008 Data System Working Group Aura
Science Meeting Columbia, MD
2Agenda
- Project Overview
- Aura Data System
- SIPS Hardware
- Software Data Flow
- Science Products
- V2.2 Releases
- GMAO Data
- V3 Development
- Quad-core
- Summary
3Project Overview
EOS MLS - Microwave Limb Sounder
- Salient Features
- One of 4 instruments on the EOS Aura spacecraft
- Launch date July 15, 2004
- Operational life 5 years
- Dr. Nathaniel Livesey (JPL) Principal
Investigator - Thermal emission from 0 to 100 km altitude in
five millimeter and submillimeter wavelength
bands - All measurements are performed simultaneously and
continuously, both day and night, and can be made
even in the presence of dense cirrus clouds and
stratospheric aerosols - Science
- Improve understanding and assessment of
- Stratospheric ozone depletion and chemistry
- Upper tropospheric ozone distribution and
chemistry - Climate change and variability
- Global distribution of O3, ClO, H2O, SO2, CO,
N2O, HNO3, HCl, HOCl, BrO, OH, HO2, HCN,
temperature, geopotential height, and cirrus ice
4Aura Data System
5Hardware Diagram
6Software Data Flow
ML0SCI1, 2
AURGBAD1
AURATTRH
AUREPHMH
ML0ENG1,2,3,4,5,6
Level 0
Level 1 Products ML1RADD, ML1RADG, ML1RADT, ML1OA
MLS Level 1 PGE
LeapSec, UTCPole
Level 1
Level 2 Geophysical data ML2BRO, ML2CLO, ML2CO,
ML2GPH, ML2H2O, ML2HCL, ML2HCN, ML2HNO3, ML2HO2,
ML2HOCL, ML2IWC, ML2N2O, ML2O3, ML2OH, ML2RHI,
ML2SO2, ML2T, ML2CH3CN
MLS Level 2 PGE
GMAO data
Level 2 Auxiliary data ML2DGM, ML2DGG
Level 2
ML0SCI1, 2 _NRT
AURGBAD1_NRT
AURATTR_NRT
AUREPHM_NRT
ML0ENG1,2,3,4,5,6 _NRT
Level 0
NRT Products ML2T_NRT, ML2O3_NRT
MLS NRT PGE
LeapSec, UTCPole
NRT
7Science Products
- Level 1 Calibrated radiances
- Count 4 granules/day
- Size _at_ 3.4 GB/day
- Level 2 Geophysical parameters as swaths
- BrO, ClO, CO, GPH, H2O, HCl, HCN, HNO3, HO2,
HOCl, IWC, N2O, O3, OH, RHI, SO2, T, CH3CN - Count 18 granules/day
- Size 1 to 2 MB each
- Level 2 Miscellaneous parameters including
clouds - Count 3 granules/day
- Size 725 MB/day
- Near Real Time Products Temperature and Ozone
- Count up to 96 x 2 granules/day
- Size lt200 KB/granule
- All output products are in HDF5 or HDF-EOS5
8V2.2 Releases
- V2.2 is the validated version of MLS Level 1
and 2 software - NRT is based on this release
- V2.20 initial release came in late 2006
- Took about 10,000 CPU-hours/data-day on a 3.06
GHz processor - V2.21 De-convolve DACS radiances in Level 1
- Began processing in March 2007
- V2.22 increased performance without change of
quality - Realized nearly 100 gain in performance dropping
to about 5,000 CPU-hours/data-day on a 3.06 GHz
processor - Gains were realized with code changes to
optimized performance and change of compiler to
Intel - Began in December 2007
- V2.23 addresses problems due to Aura FMU
anomalies - This version used for data dates since Dec 4, 2007
9V2.2 Calendar
10GMAO Data
- MLS Level 2 products are tied to GMAO output
- GEOS5 5.01 2004 August 8 (d221) 2007 November
30 (d334) - GEOS5 5.10 2007 December 1 (d335) 2008
September 14 (d258) - GEOS5 5.20 2008 September 15 (d259) forward
11NRT
- MLS has a NRT processor to produce two products
every 15 minutes - Temperature and Ozone
- MLS uses a fast 1D version of the forward model
- Working on a second derivative version to give
more accuracy - Typically, NRT Level 0 data is available within 2
hours and Level 2 data is available within 3
hours - Data is now publically available via GES-DISC
- NRT Data Quality Description Document is released
- Thanks to GES-DISC for their effort in producing
this data in a timely manner
12V3 Development
- V3 development is on-going and expect to have in
production in late spring/early summer of 2009 - Expected features include
- Forward Model improvements
- Investigate instrument related artifacts such as
antenna transmission, filter shifts, and gain
compression - Investigate line shape absorption model
- Study cloud radiation scattering effects
- Product improvements
- Extend measurements vertically
- Increase vertical grid for O3
- Address biases
- Improve products in clouds
- New products CH3Cl, CH3OH, CH3COCH3, Ice Water
Path
13Quad-Core Performance
- MLS has explored the use of Intel quad-core
- MLS ran our current Level 2 with no
re-compilation - It ran nearly twice as fast on each processor
- Each quad-core has 4 times as many processors as
our current architecture - We think we may be able to realize re-processing
5 years of MLS data in about 7 months with 154
nodes - Each node has two processors and 16 GB of memory
- With 16 GB memory per board (memory is
expensive), we have seen significant increase in
throughput - It may not be backward compatible with our
current architecture
14Summary
- EOS MLS is healthy
- MLS is producing validated geophysical products
- MLS is producing Temperature and O3 for Near Real
Time - MLS tries to stay in sync with GMAO output
- MLS is working toward V3
- MLS hopes to leverage performance improvements
from Intel