Radiation Belt Tools and Climatology Eric A. Kihn – NOAA/NGDC Paul O’Brien- Aerospace Robert Weigel – GMU Completing the Data Environment
Radiation Belt Tools and Climatology
Eric A. Kihn – NOAA/NGDC
Paul O’Brien- Aerospace
Robert Weigel – GMU
Completing the Data Environment
State of the Data Environment• Observational data is available, but often
scattered, minimally documented and difficult to access
• Working with data often involves directly securing a PI’s time or a minor research project
• No sense of community support or access for those outside the field
Science Data Stewardship
•A focus on reaching a broader customer base•An effort to reduce redundant functions on the data•An effort to improve understandability through metadata•A new focus on machine based access support multiple community based front-ends
Example Data Set: POES- MEPAD
•Data available at poes.ngdc.noaa.gov (78-08)•Data in POES binary•Has known contamination issues•Preview (QC) plots in linear time•16 sec avg data as CDF or ASCII
Improved Data Product
•Data available at poes.ngdc.noaa.gov N15 and later•Data in NetCDF•The cross channel contamination has been removed (Green)•Preview plots in L-shell and include Auroral Oval•Full time resolution data •Full metadata record in SPASE format
Data Matrix
• 1.1 Reanalysis • 1.2 AMPTE • 1.3 SAMPEX • 1.4 GOES • 1.5 POES • 1.6 METOP • 1.7 HEO • 1.8 GPS • 1.9 LANL GEO • 1.10 Polar • 1.11 CRRES • 1.12 Akebono • 1.13 SCATHA • 1.14 ICO • 1.15 S3-3 • 1.16 OV3-3
Details: http://virbo.org/wiki
Climatology vs. Reanalysis
• Gives you min/max/mean
• Is derived from direct observations
• Is useful for quick look-up of environmental specs
• Doesn’t contain the correlations between observables
• Gives you “a” state representation
• Is derived observation plus model
• Is useful for extracting scenarios
• Represents the physical correlations and boundaries
Introduction to Reanalysis
• Scientists around the world use the reanalysis data for:
– Climate studies
– Seasonal climate prediction
– Climate variability studies
– Initial/boundary conditions for regional/sub-grid-scale models
– Diagnostic studies
– Verification of climate models
– Testbed for operational models
• The US atmospheric science community produces a standardized ‘reanalysis’ (via NCEP and NCAR)
• The reanalysis is built by going back as far as the data allows and running a consistent standard data assimilative physics-based global analysis model
• The reanalysis provides numerous climate and weather data for the entire globe on a standard grid. The reanalysis is run after the fact, when all data are available.
July 29, 2004 Reanalysis:Air Temp at Sea Level (K)
Figure courtesy US National Climate Reanalysis Project
Space Weather Analysis
IMFKpDst
10.7 cm FluxHPI
MagnetometerGOES
AMIE
GITM
SIMMHigh Lat Elec
Kp
Init Conditions
SWRDATA
TEC, FoF2,Neutral Winds
Magnetic, Electric Potential, Etc.
Particle Data
New HPI database (DMSP, NOAA)New 100 + magnetometer database.
210 MM, Canopus, Tromso, Greenland, Image, etc..Complete IMF RecordAMIE Runs @ 1.0 minute (1989-2003)GITM Runs (1991-2002)SIMM runs (1991-2001)
Pros and Cons of ReanalysisPros
• The final product, a “Standard Solar Cycle” is conceptually simpler than a model that attempts to statistically characterize the temporal dependence
• Reanalysis can be stored on any coordinate system (even time-alt-lat-lon!)
• Specifications for different domains with their own natural coordinates can be combined on a single, common coordinate system (again, e.g., time-alt-lat-lon)
• Reanalysis captures real events rather than simulated ones, thus capturing realistic temporal correlations (especially useful for determining the frequency and duration of an effect)
Cons• Probably a lot more work than
mean/worst case flux maps• Smooths out spatial variations
(artificially increases spatial correlations)
• May not accurately capture tails of distributions (we must be careful about this)
• (Much) larger database– This is much less of a
concern now
What makes it “Reanalysis”• Part of the fitting procedure is to determine the best estimate for the state x of the system conditioned on minimizing the error between the observations y and the estimates of those observations . The measurement matrix H relates the fluxes to the
observations (which are typically count rates in a detector)
• It is important to note that in reanalysis we do NOT try to convert the observations into fluxes or phase space densities. Rather, we use the instrument response function to “predict” what the instrument would measure given the state x and penalize that state x for any deviation between those “predictions” and the observations. Knowledge of the instrument response (and its uncertainty) becomes paramount.
• The observation penalty function (pe) is multiplied by another function that penalizes deviations from the output of a statistical p(x) or physics-based model for x:
Example Statistical Reanalysis Results
Energetic Electron Statistical ReanalysisInner and Outer belt electron flux from 100 keV to 7 MeVDerived from a static model of statistical variationThis reanalysis covers a full solar cycleIt was constrained with HEO-1 and HEO-3 dataPrior to the launch of HEO-3, the specification is essentially useless at this energy (703 keV)There are also spectral features (e.g., bumps) that don’t appear to be realistic
@ GPS
Coming Work in this Area
• GEM Focus Group -Space Radiation Climatology (2006 - 2011, P. O'Brien and G. Reeves) -([email protected], [email protected])
• ONERA – Salammbo model has data assimilative models for GPS and GEO satellites
• LANL-DREAM project is pursuing a more ambitious model that couples the radiation belt into a global model that includes the ring current, plasmasphere and convection electric fields.
Quality Control
• In activities like a reanalysis a lot of issues fall out
• New tools that more easily identify and retrieve satellite conjunctions will
• Better metadata and metadata accessibility should document instrument temporal changes
• Needs to be an on going community coordinated effort.
Virtual Radiation Belt Observatory (ViRBO)
Data
Models
Reanalysis
Data
Data
Models
Software
Documents
End User
ViR
BO
AP
I
Custom Interface
Commercial Interface
Conclusions• The new data stewardship paradigm will mean a
fundamental shift in the way research is done and provide many opportunities to operations.
• The tremors are already past the data center level profoundly effecting the center missions.
• “Most researchers are accustomed to studying a
relatively small data set for a long time, using statistical models to tease out patterns. At some fundamental level that paradigm has broken down.” – Nature June, 1999