The San Simeon and Parkfield Earthquakes

1
The San Simeon and Parkfield Earthquakes

• What the CISN did well
• (and could have been done better)

2
M3 quake ? M6 quake

• Big quakes
• Have extended source regions
• Are infrequent
• Exercise software in unexpected ways
• Shine spotlights on systems deficiencies
• Overwhelm humans unaccustomed to response
  pressures and demand for information
• Rarely happen when convenient

3
San SimeonMw6.5 Earthquake

• Quake occurred December 22, 2003 at 111556 AM
  (PST)
• Monday
• Staff gone for vacation
• Instrumentation was (and still is) sparse
• Instrumentation was (and still is) analog

Quake occurred on edge of NC/SC networks
4
12/22/2003 191556
000000
Rupture initiates
5
?
12/22/2003 191620
000024
Quick Look Hypocenter
6
12/22/2003 192025
000429
Final hypocenter Md 5.6
7
12/22/2003 192058
000502
ML 6.4
8
12/22/2003 192240
000644
Mechanism Mw 6.5
9
12/22/2003 192413
000817
1st ShakeMap - ML 6.4
10
12/22/2003 193828
002232
ShakeMap update - Mw 6.5
11
San Simeon Human Time-line
Time Elapsed (GMT) hrmnsc 193200 001604 S
eismologist review and aftershock
probabilities 200000 004404 OES
conference call with CISN scientists 201600 010
004 First HAZUS runs completed 203000 011404
Release of Internet Quick Report 215400 023804
Seismologist review of finite source
12
12/22/2003 203000
011404
1st Internet Quick Report
13
023804
14
12/22/2003 221400
025804
ShakeMap update
15
12/22/2003 233300
041704
ShakeMap update - line source
16
12/23/2003 023400
071804
ShakeMap update - Templeton data
17
12/24/2003 002800
291204
ShakeMap update - Cambria data
18
12/23/2003 173400
221804
Special Earthquake Report
19
J. Fenton
20
So far, so good.Heres what could have gone
better

• Technical issues
• Competing EqintheNews pages on USGS servers in
  Golden, CO and Menlo Park, CA (fixed)
• No ability to update aftershock probabilities
  (now can be done manually)
• Clients inundated with automated email/pager
  (potentially solvable)
• Akamai/Reston firewall problems blocked recenteqs
  access (fixed)
• CIT ShakeMap data exchange broken (fixed)
• Dedicated com links from Parkfield broke (fixed)

21

• Human Response
• Inconsistent statements on mapped fault and name
  of quake on Day 1
• No communications with ShakeMap users about
  ShakeMap updates
• Failure to exercise existing response procedures
• SC "duty seismologist" for ShakeMap unknown in NC
• 2 days to establish an internal scientific email
  distribution list and ftp site
• No 7 X 24 contact for USGS networking/security
  staff. Even after contact, fix took almost 2
  weeks.

22
3. Data Quality
Distribution of ShakeMap stations

• Limited station coverage and strong motion data
  in region compromised ShakeMap quality and science

RT Non-RT
23
4. Statewide pick exchange and association

• Prototype operational
• better locations
• detected more earthquakes

24
ParkfieldOverview

• Quake occurred Tuesday, September 28, 2004 at
  111556 AM (PST)
• Key personnel in St. Louis for National
  Earthquake Conference
• Quake occurred within densely instrumented
  portion of NC network
• SM data mostly non-RT

25
Parkfield time-line

• Earthquake Information UTC Time Elapsed time
• hrmnsc
• Origin Time (OT) 09/28 171524 000000
• Quick Look hypocenter 09/28
  171555 000031
• Final hypocenter Md 5.8 09/28
  171957 000433
• First Motion mechanism 09/28
  172010 000446
• Local Magnitude ML 6.0
  09/28 172030 000506
• 1st ShakeMap completed 09/28
  172116 000552
• Moment Tensor Solution Mw 5.9 09/28
  172223 000659
• 2nd ShakeMap completed 09/28
  172808 000744
• MS relocated by duty seismologist 09/28
  1746-- 0031--
• Analyst review of moment tensor 09/28
  1747-- 0032--
• ShakeMap update due to relocation 09/28
  1753-- 0038--
• Finite fault line source 09/20 1754-- 0039
• Probability statement released for ML5.9 09/28
  1800-- 0045--
• ShakeMap update with manual finite fault 09/28
  1820-- 0105--
• Update to Mw6.0 probability revision 09/28
  1823-- 0108--
• ShakeMap update with 6 stations 09/28
  2118-- 0403--
• Earthquake Report at cisn.org 09/28
  2145-- 0430--

Automatic
OES/CISN exchange
26
Locations almost perfect
27
100 km distance weighting

• 28 SEP 2004, 1715 SEQUENCE NO. 1, ID NO.
  51147892
• 
  ADJUSTMENTS (KM)
• I ORIGIN LAT N LON W Z NWR RMS
  DT DLAT DLON DZ RR NF MOD
• 1 23.98 35 50.15 120 21.53 5.00 250 1.59
  0.600 -6.516 7.195 0.000 9.707 3 PGH
• 2 24.58 35 46.63 120 26.32 5.00 250 1.00
  -0.015 -0.476 0.404 0.000 0.624 3 PGH
• FREE DEPTH
• 3 24.57 35 46.38 120 26.58 5.00 250 1.00
  -0.453 0.106 0.273 -4.461 4.470 4 PGH
• BEGIN DISTANCE WEIGHTING
• BEGIN RESIDUAL WEIGHTING
• DEPTH ADJUSTMENT DAMPED
• 4 24.11 35 46.43 120 26.77 0.54 49 0.75
  9.804 3.234 -4.745 15.70216.719 4 PGH
• RMS INCREASE - MOVE HYPO 0.60 BACK
• 5 33.92 35 48.18 120 23.61 16.24 51 9.97
  -5.882 -1.940 2.847 -9.42110.032 4 PGH
• RMS INCREASE - MOVE HYPO 0.60 BACK
• 6 28.03 35 47.13 120 25.50 6.82 51 3.82
  -2.353 -0.776 1.139 -3.769 4.013 4 PGH
• RMS INCREASE - MOVE HYPO 0.60 BACK
• 7 25.68 35 46.71 120 26.26 3.05 51 1.69
  -0.941 -0.310 0.456 -1.507 1.605 4 PGH
• RMS INCREASE - MOVE HYPO 0.60 BACK
• 8 24.74 35 46.55 120 26.56 1.54 51 0.98
  -0.376 -0.124 0.182 -0.603 0.642 4 PGH

28
24 hours
29
10 days later
30
Web traffic

• Request or hit count peak of 630,000/min
  (10,500/sec)
• 300,000 different IP addresses on day 1
• 100GB transferred on day 1 (typically 8 GB)

31
Did You Feel It?

• 7,000 questionnaires submitted for this event in
  the first hour
• 14,174 total

32
430
33
841
34
Parkfield Issues

• Technical
• Mainshock and two M5 aftershocks initially
  mislocated (fixed)
• Strong motion data exchange from CGS hung
  (fixed?)
• Reviewed Mw update had to be forced (not fixed)
• Incorrect aftershock M5 magnitudes (fixed?)
• Human
• Duty seismologist failed to follow procedures and
  aftershock message was delayed
• Staff overload because key individuals in St
  Louis
• Data
• Most strong-motion data non-RT and recorded on
  film

35
Summary

• CISN delivered rapid and reliable information for
  San Simeon and Parkfield quakes.
• Technical glitches were addressable
• Human glitches challenging
• Limited instrumentation and data quality remain a
  significant impediment