The Yin and Yang of Best Track Data: User "needs" and Producer "wants" Richard J. Murnane RPI/BIOS and Baseline Management Company, Inc. 7 May, 2009 IBTrACS.

The Yin and Yang of Best Track Data:

User "needs" and Producer "wants"

The Yin and Yang of Best Track Data:

User "needs" and Producer "wants"

Richard J. MurnaneRichard J. MurnaneRPI/BIOS and Baseline Management Company, Inc.RPI/BIOS and Baseline Management Company, Inc.

7 May, 20097 May, 2009IBTrACS WorkshopIBTrACS Workshop

Asheville, NCAsheville, NC

Overview• Motivation: A (re)insurer’s view of hurricane science

• RPI, (re)insurance, and extreme events

• Examples of RPI-funded research related to best-track data

• Another example: offshore risk

• Previous workshop on western North Pacific best-track data

2001 West Pacific Tracks And Typhoon Lingling

Lingling Wind Speed EstimatesAgency Speed (kts) Category

Maximum wind speed (satellite obs)

China 115 Cat 4JTWC 115 Cat 4Hong Kong 109 Cat 3JMA 97 Cat 3

Just before landfall (satellite obs)China 91 Cat 2JTWC 95 Cat 2Hong Kong 75 Cat 1JMA 57 TS

Measured windVietnam 71(?) Cat 1

“… future warming may lead to an upward trend in tropical cyclone destructive potential, and – taking into account an increasing coastal population – a substantial increase in hurricane-related losses in the twenty-first century.”

Changes In Hurricane Power?

Year

Se

a S

urf

ace

Te

mp

era

ture

Po

we

r D

issi

pat

ion

Ind

ex

K. Emanuel, Nature, 2005.

Changes In Intense Hurricanes?

“… global data indicate a 30-year trend toward more frequent and intense hurricanes, …”

Webster et al., Science, 2005.

Or, No Change?

“Subjective measurements and variable procedures make existing tropical cyclone databases insufficiently reliable to detect trends in the frequency of extreme cyclones.”

Landsea et al., Science, 2006.

“… the increase of [vertical wind shear] has been historically associated with diminished hurricane activity and intensity. A suite of state-of-the-art global climate model[s] project… [s]ubstantial increases in tropical Atlantic and East Pacific shear …”

Future Unfavorable Conditions?

Vecchi and Soden, GRL, 2007.

Upward Trend In Strongest Storms?

“... possible trends … are less obvious, owing to the unreliability and incompleteness of the observational record... Here we overcome these two limitations by examining trends … from homogeneous data derived from … satellite records. We find significant upward trends for wind speed quantiles above the 70th percentile…”

Elsner et al., Nature, 2008

Yin and Yang

• Climate Signal– Yes or No?

State Of Knowledge

IntergovernmentalPanel onClimateChange

Attribution And Projection

CCSP, 2008

Catastrophe Risk Model

physical damagerepair costs

Damage

terms of coverage

Insured Loss

ProbabilityLocation

Intensity, Wave HeightDuration

HazardLocation

ConstructionAge

Building Code

Exposure

Yin and Yang

• Science– Can delay decisions

for further study– Should be better

than previous attempts

– 95% certainty

• Private Sector– Must make a

decision (now)– Good enough is

sufficient– 51% certainty will

drive a decision


• RPI, (re)insurance, and extreme events




RPI Corporate Sponsors• XL Re Ltd.

• PartnerRe

• Amlin Underwriting

• Renaissance Reinsurance Corporation

• Axis Specialty

• Nephila Capital

• State Farm

• Aspen Insurance

• Risk Management Solutions

• FlagstoneRe

Top 40 Property Cat Losses 1970-2007

Onshore Katrina/Rita LossesOffshore

Losses total$308 billion in2007 dollars

Swiss Re Sigma, 1/2008

Top 30 For Victims (1970-2006)

Number of Victims

0

100,000

200,000

300,000

400,000

500,000

600,000

700,000

800,000

900,000

Weather Earthquake Man-Made Volcano

Hazard


2007 Non-life Premium Volume

Swiss Re, Sigma 3/2008

The US and Europe account for ~80% of global premium:~$1.7 trillion

Top 30 For Victims (1970-2006)

Number of Victims

0

100000

200000

300000

400000

500000

600000

700000

800000

900000

Location of Catastrophe


Yin and Yang• Science

– Information from any basin is valuable, Pacific of potentially greater interest than Atlantic because of sample size

– ENSO or climate a major driver

– Focus on over-ocean behavior

– More responsive (concerned?) to human or environmental factors

• Private Sector– Mainly interested in

the Atlantic (but Pacific is of growing import)

– ENSO of interest, but a single storm in any year drives losses

– Focus on over-land characteristics

– More concerned with monetary factors


• RPI, (re)insurance, and why they’re not interested in all extreme events




Impact Of Storm Size

• Satellite data:– Available on a global

basis– Records back to

1970s– Untapped potential

information

2005 Hurricane Dennis2004 Hurricane Ivan

Analyzed Vmax 100kts7 nm from center

Insured loss of$1.1 Billion

Analyzed Vmax 95kts20 nm from center

Insured loss of$7.1 BillionH*Wind analyses from HRD

Extended Best-Track Data

• 1997 RPI workshop: “Wind Field Dynamics of Landfalling Tropical Cyclones”

• Mark DeMaria said there were boxes of observations sitting in hallway

• RPI funded effort to digitize data: Extended Best Track data set

Wind Radii From Satellite Observations

Lower latitude Higher latitude

Kossin, 2005Kossin, 2005

UW/NCDC Reanalysis For PDIAtlantic NE Pacific NW Pacific

Global

Kossin et al., 2007

Formation and Tracks of 1970-2003 European Impacts

Evans and Hart, 2005






Offshore Industry In Gulf Of Mexico

Offshore structures: >4000Length of pipeline: >56,000 kmProperty Value: ~$150 Billion

MMS, 2008

Oil and gas pipelines with diameters ≥ 20 inches

Hurricane Ike (2008)

Losses from : http://www.gccapitalideas.com/2009/01/21/synopses-of-significant-tropical-cyclones-in-2008/

Cat Model Date Onshore Losses Offshore Losses(Billions $) (Billions $)

AIR Sept. 13 8 - 12 0.6 - 1.3December 10 - 15 1 - 2

EQECAT Sept. 13 8 - 18Sept. 19 8 - 12 4 - 6

RMS Sept. 14 6 - 16Sept. 17 7 - 12Oct. 24 13 - 21 1 - 3

To a great extent, offshore losses driven by storm wind field, motion, and track because of wave damage

• Environmental issues covered (in 3.5 pages) are:– Deep-sea currents– Deepwater shipwrecks (finding them is

a benefit)– Environmental impacts, mainly on

biology

• The word:– “hurricane” occurs 5 times, mainly in

conjunction with an explanation of a drop in production

– “weather” occurs 2 times (in a single paragraph)

– “climate” does not rate a single mention…

U.S. Department of the Interior, Minerals Management Service, Gulf of Mexico OCS Region, 102 pages, New Orleans, May 2008

Minerals Management Service Report






2001 RPI Workshop

• Potential Development of a Unified Northwestern Pacific (NWPAC) Tropical Cyclone Best-Track Data Set – http://w3.bios.edu/rpi/public/meetings/2001/nov01/agenda.htm

– Chris Landsea, Colin McAdie, Mark DeMaria, Chip Guard, Tatsuo Ueno, Chris Cantrell, Shangyao Nong

Reinsurer’s Perspective:

Areas Or Countries Of Most Interest

• In general, all countries exposed to tropical cyclones, but a rough order of importance is:– Japan– South Korea– Taiwan, Philippines– Hong Kong– Mainland China

Reinsurer’s Perspective:Type Of Desired Information

• Two levels of information– Level 1: Mainly basic information

for calculating gradient winds• Location on 3-hour basis• Central pressure• Radius of maximum winds• Maximum wind• Forward Movement

Reinsurer’s Perspective:Type Of Desired Information

• Two levels of information– Level 2: Full wind field and

precipitation data, e.g.,• Radii of 34 and 50 knot winds• Sustained winds versus peak gusts• Rainfall rate

Reinsurer’s Perspective:Data Format And Availability

• Data format relatively unimportant, as long as it is standardized

• Comma-delimited ASCII probably most practical

• On-line access preferred

http://w3.bios.edu/rpi/public/meetings/2001/nov01/agenda.htm

Reinsurer’s Perspective:Use Of “Real-Time” Wind Data

• With initial storm landfall try to overlap wind field data to “exposure” information to estimate expected losses

• At other times use proprietary models that have a version of integrated wind fields

WMO Report Data

• CI and T numbers• Confidence in center

position• Max gust and quality

code• Rmax• Wind radii: gale force

and optional wind speed in 4 quadrants

The Yin and Yang of Best Track Data: User "needs" and Producer "wants" Richard J. Murnane RPI/BIOS and Baseline Management Company, Inc. 7 May, 2009 IBTrACS.

Documents