TIC Working Group E, Evolutionary System Architecture

1
TIC Working Group E Evolutionary System
Architecture
Walter Arabasz David Oppenheimer
March 3, 2005
2
Working Group Members...

• Walter Arabasz, Chair (Univ. of Utah, NIC TIC)
• Glenn Biasi (Univ of Nevada, Reno)
• Ray Buland (USGS Golden NEIC)
• Art Lerner-Lam (Lamont-Doherty Earth Observatory
  IRIS)
• Phil Maechling (Univ. of Southern California
  SCEC)
• Tom Murray (USGS Anchorage AVO)
• David Oppenheimer (USGS Menlo Park, NIC CISN)
• Rick Schult (Air Force Research Lab, Hanscomb
  AFB)
• Tony Shakal (California Geological
  Survey/Strong-Motion
  Instrumentation Program CISN)
• Mitch Withers (Univ of Memphis NIC)

3
Charge

• Define an evolutionary path for transforming
  existing elements of ANSS into a functional
  nationwide systemwith emphasis on steps that can
  be taken in the near term (1-3 yrs), based on
  realistic ANSS funding projections
• Clarify key system performance goals relevant to
  system design and characterize where we are
  now
• Account for geopolitical realities as well as
  abstract ideals in designing an ANSS system
  architecture

4
Presentation Outline

• Guiding Principles (Baldrige and Road Map)
• Review key system performance goals
• Characterize where we are now
• Discuss (in detail) 3 architectures for an ANSS
  system
• Recommendations

5
Baldrige National Quality Program

• A NIST-sponsored program for a systems-level
  approach to organizational excellence
• Provides assessment, self-improvement, and
  planning tools
• Leadership
• Strategic Planning
• Customer (and Market) Focus
• Measurement, Analysis, and Knowledge
• Human Resource Focus
• Process Management
• Organizational Performance Results

6
Road Map for Partnership
How do we reconcile state/local
ownership, investment in, and ongoing support of
significant infrastructure for seismic
monitoring with the prescriptions of
ANSS decision makers?
7
We need to persuade network operators (and their
sponsors) to move ahead toward a better-designed
nationwide system that offers a win-win deal for
both individual networks and the system.
8
Key System Goals

• Rapid Parametric Information
• Data Exchange
• Information Distribution
• Quality Control
• Security
• Public Archive
• One earthquake, one report
• Reliability

9
Where We Are Now

• 20 Questions distributed prior to WG-A, but
  compared against proposed standards
• Report discusses survey and provides link to all
  responses
• No seismic network meets proposed standards
• Effort will be substantial to meet proposed
  standards

10
Where-We-Are Now Findings

• Standardization of algorithms lacking
• Need for reconciliation of multiple reports of
  earthquakes
• Limited centralized waveform archiving
• No standardized error estimates
• Inadequate metadata
• Uneven exchange of waveform data between networks
• Little strong motion processing

11
Where-We-Are Now Findings

• No uniformity of magnitude calculation
• Moment tensor calculations produced only by AEIC,
  CISN, and NEIC
• ShakeMaps produced only by PNSN, CISN, Utah, and
  Nevada
• Parametric data publicly available only from
  NEIC, AEIC, CISN, and Utah

12
Consider 3 ANSS Architectures

• Decentralized
• Processing occurs at regional centers
• Product conflict resolved nationally
• Backed up by national facility
• TIC Plan
• Like Decentralized but one center per region
• Centralized
• All processing at a national facility or IPS
• Raw data (waveforms, picks) forwarded from data
  concentrators
• Always authoritative, but backed up by regions

13
Decentralized Processing
WEB
EQalert
OFR 02-92 Nomenclature
Info outlet
Data processing
ANSS Central Site
Concentrator
Waveforms
Products
Archive(s)
Regional Seismic Networks
EOC OES
WEB
Stations
EQalert
14
Pros
Cons

• Similar to current situation
• Robust since data close to processing
• Autonomy fosters local solutions
• Primary role justifies local funding
• Local knowledge utilized
• Regional data sharing sufficient to monitor

• Facilities at risk from earthquakes
• Difficult to standardize data exchange
• Rules required to resolve authoritative
  information
• Expensive to staff 7X24
• Difficult to integrate global data sets into
  local archive
• Duplication of efforts potentially wasteful

15
TIC Plan
WEB
EQalert
OFR 02-92 Nomenclature
Info outlet
ANSS Central Site
Data processing
Concentrator
Waveforms
Products
Archive(s)
Regional Centers
EOC OES
WEB
Stations
Subregional Seismic Networks
EQalert
Stations
16
Pros
Cons

• Similar to Decentralized model
• Processing could be performed in areas of lower
  seismic hazard
• Fewer units decreases complexity of system

• Similar to Decentralized model
• Potentially expensive to establish new regional
  centers, and given current level of funding,
  unlikely to receive much support
• Uneven work loads from region to region

17
Integrated Processing Service
WEB
EQalert
OFR 02-92 Nomenclature
Info outlet
ANSS Archive IRIS, NCEDC, and SCECDC
Data processing
continuous?
IPS
Concentrator
Waveforms
Products
Waveforms or picks snippets
products
EOC OES
FEMA, NOAA
RSNs
NEIC
WEB
Stations
EQalert
18
Pros
Cons

• Simplifies standardization and delivery
• Comprehensive view of earthquake
• Integrates global data for large US quakes
• Eliminates conflicting reports
• Minimizes 7X24 cost
• IPS could be located in area of minimal seismic
  risk
• Experienced staff respond
• Local scientists unburdened from technical
  response during crises
• All products go into a central DBMS
• Continuous waveform archive possible
• Distribution of RT waveforms to RD groups
• Single connection for Earthscope USArray and PBO
  data streams

• Data less robust due to long paths
• Loss of local knowledge
• Requires methodologies for local/regional/global
• Regional identity (funding?) diminished
• Local incentive to invent diminished
• Data analyst motivation diminished
• Could take years to develop
• Single point of algorithm, hardware, and
  distribution failure
• Transfer of software back to regional networks
  disruptive
• Full waveform exchange costly over DTS impacts
  campus traffic if over Internet

19
WG-E Recommendations

• Software management group (SMG) (person?) should
• write guidelines for ANSS software oversight with
  TIC/NIC review/approval
• develop specifications for next generation of
  ANSS software
• include cost estimates and milestones
• address regional and global needs
• complete work by 10/31/2005 (?)

20
Recommendations continued

• Software should
• be open source if possible
• evaluated in its full context of development,
  ownership, and maintenance
• permit centralized, decentralized, and hybrid
  modes of operation. One size does not fit all.

21
Recommendations continued

• TIC should
• allow/invite comment on specifications
• have authority to modify specifications based on
  comments
• ensure specifications have broad political and
  technical support across ANSS

22
Recommendations continued

• ANSS management should
• review document in early FY06 allocate funding
• Invite universities to participate in development
  of software projects
• Conduct a benefit-cost analysis in FY06 or FY07
• Adopt Baldrige National Quality Program

23
Recommendations continued

• ANSS management and principal stakeholders should
  develop a standard MOA that
• defines how partners will participate in ANSS
• considers political issues (regional/state/local,
  centralized/decentralized/hybrid, Road Map for
  Partnership)
• defines performance standards to be met

24
Recommendations (finally..)

• Regarding OFR 02-92, WGE
• Abandon concept of one primary center per ANSS
  region

25
Closing thoughts

• WGE was unable to reach full consensus on the end
  state. Geopolitical realities are clearly a
  paramount challenge
• Network operators are justifiably concerned about
  self-preservation
• We have the opportunity to be visionary