27
Slide 1 RMetSoc Hurricanes March 2006 How Well Forecast Were the How Well Forecast Were the 2004 and 2005 Atlantic and 2004 and 2005 Atlantic and U.S. Hurricane Seasons? U.S. Hurricane Seasons? Mark Saunders and Adam Lea Mark Saunders and Adam Lea Department of Space and Climate Physics Department of Space and Climate Physics Benfield Hazard Research Centre Benfield Hazard Research Centre University College London University College London Royal Meteorological Society Meeting Royal Meteorological Society Meeting Hurricanes Hurricanes 15th March 2006 15th March 2006

How Well Forecast Were the 2004 and 2005 Atlantic and U.S. Hurricane Seasons? Mark Saunders and Adam Lea Department of Space and Climate Physics Benfield

Embed Size (px)

Citation preview

Page 1: How Well Forecast Were the 2004 and 2005 Atlantic and U.S. Hurricane Seasons? Mark Saunders and Adam Lea Department of Space and Climate Physics Benfield

Slide 1RMetSoc Hurricanes March 2006

How Well Forecast Were the 2004 How Well Forecast Were the 2004 and 2005 Atlantic and U.S. and 2005 Atlantic and U.S.

Hurricane Seasons? Hurricane Seasons?

Mark Saunders and Adam LeaMark Saunders and Adam Lea

Department of Space and Climate PhysicsDepartment of Space and Climate PhysicsBenfield Hazard Research CentreBenfield Hazard Research Centre

University College LondonUniversity College London

Royal Meteorological Society Meeting Royal Meteorological Society Meeting HurricanesHurricanes

15th March 200615th March 2006

Page 2: How Well Forecast Were the 2004 and 2005 Atlantic and U.S. Hurricane Seasons? Mark Saunders and Adam Lea Department of Space and Climate Physics Benfield

RMetSoc Hurricanes March 2006 Slide 2

2004/5 Hurricane Seasons2004/5 Hurricane Seasons

• Rank as the most Rank as the most active and damaging active and damaging consecutive hurricane consecutive hurricane years on record. years on record.

• SevenSeven intense intense hurricane landfalls on hurricane landfalls on the U.S. (norm is one).the U.S. (norm is one).

• Estimated total Estimated total damage bill damage bill approaching approaching US $ 200 US $ 200 bnbn (norm is US $ 10bn). (norm is US $ 10bn).

11:00 UT 29 August 200511:00 UT 29 August 2005 (Courtesy NOAA)(Courtesy NOAA)

Page 3: How Well Forecast Were the 2004 and 2005 Atlantic and U.S. Hurricane Seasons? Mark Saunders and Adam Lea Department of Space and Climate Physics Benfield

Slide 3RMetSoc Hurricanes March 2006

StructureStructure

1.1. How unusual were 2004 and 2005? How unusual were 2004 and 2005?

2.2. Reasons for high Atlantic and U.S. Reasons for high Atlantic and U.S. hurricane activity in 2004 and 2005. hurricane activity in 2004 and 2005.

3.3. Seasonal Forecast Comparison Seasonal Forecast Comparison (a)(a) Atlantic basin hurricane activity. Atlantic basin hurricane activity. (b)(b) U.S. landfalling hurricane activity. U.S. landfalling hurricane activity.

4.4. Forecast Business Application. Forecast Business Application.

5.5. Conclusions.Conclusions.

Page 4: How Well Forecast Were the 2004 and 2005 Atlantic and U.S. Hurricane Seasons? Mark Saunders and Adam Lea Department of Space and Climate Physics Benfield

RMetSoc Hurricanes March 2006 Slide 4

2005 Hurricane 2005 Hurricane SeasonSeason

Most active and Most active and damaging on record.damaging on record.

Most tropical storms: Most tropical storms: 2727

Most hurricanes: Most hurricanes: 1515

Most Cat 5 hurricanes: Most Cat 5 hurricanes: 33

Most major hurricanes to Most major hurricanes to strike the U.S.: strike the U.S.: 44

Highest US hurricane Highest US hurricane insured damage: insured damage: ~ US $ ~ US $ 50bn.50bn.

11

Page 5: How Well Forecast Were the 2004 and 2005 Atlantic and U.S. Hurricane Seasons? Mark Saunders and Adam Lea Department of Space and Climate Physics Benfield

Slide 5RMetSoc Hurricanes March 2006

Major Gulf Hurricanes in 2004/5Major Gulf Hurricanes in 2004/5

First time since 1915/6 that four major hurricanes haveFirst time since 1915/6 that four major hurricanes havestruck the Gulf Coast in two years.struck the Gulf Coast in two years.

Page 6: How Well Forecast Were the 2004 and 2005 Atlantic and U.S. Hurricane Seasons? Mark Saunders and Adam Lea Department of Space and Climate Physics Benfield

Slide 6RMetSoc Hurricanes March 2006

Florida Damaging Storms 2004Florida Damaging Storms 2004

Courtesy of NOAA and UW-CIMSSCourtesy of NOAA and UW-CIMSS

First time since 1886 that four hurricanes haveFirst time since 1886 that four hurricanes havestruck the same state in one year. struck the same state in one year.

Page 7: How Well Forecast Were the 2004 and 2005 Atlantic and U.S. Hurricane Seasons? Mark Saunders and Adam Lea Department of Space and Climate Physics Benfield

RMetSoc Hurricanes March 2006 Slide 7

Most Costly US Insured LossesMost Costly US Insured Losses

Page 8: How Well Forecast Were the 2004 and 2005 Atlantic and U.S. Hurricane Seasons? Mark Saunders and Adam Lea Department of Space and Climate Physics Benfield

RMetSoc Hurricanes March 2006 Slide 8

2. Reasons for high Atlantic2. Reasons for high Atlantic and U.S. hurricane activity in and U.S. hurricane activity in 2004 and 2005 2004 and 2005

Page 9: How Well Forecast Were the 2004 and 2005 Atlantic and U.S. Hurricane Seasons? Mark Saunders and Adam Lea Department of Space and Climate Physics Benfield

RMetSoc Hurricanes March 2006 Slide 9

August/September 2004/5 Sea Surface August/September 2004/5 Sea Surface Temperature Anomaly (1950-2003 average)Temperature Anomaly (1950-2003 average)

+0.67°C SST anomaly in hurricane +0.67°C SST anomaly in hurricane main development regionmain development region

Page 10: How Well Forecast Were the 2004 and 2005 Atlantic and U.S. Hurricane Seasons? Mark Saunders and Adam Lea Department of Space and Climate Physics Benfield

RMetSoc Hurricanes March 2006 Slide 10

August/September 2004/5 Anomaly in Vertical August/September 2004/5 Anomaly in Vertical Wind Shear (1995-2003 average)Wind Shear (1995-2003 average)

Page 11: How Well Forecast Were the 2004 and 2005 Atlantic and U.S. Hurricane Seasons? Mark Saunders and Adam Lea Department of Space and Climate Physics Benfield

RMetSoc Hurricanes March 2006 Slide 11

August/September 2004/5 Mean Sea Level August/September 2004/5 Mean Sea Level Pressure Anomaly (1995-2003 average)Pressure Anomaly (1995-2003 average)

Page 12: How Well Forecast Were the 2004 and 2005 Atlantic and U.S. Hurricane Seasons? Mark Saunders and Adam Lea Department of Space and Climate Physics Benfield

RMetSoc Hurricanes March 2006 Slide 12

August/September 2004/5 925-400mb Height August/September 2004/5 925-400mb Height Averaged Wind Anomalies (1995-2003 average)Averaged Wind Anomalies (1995-2003 average)

Steering Steering winds winds on-shoreon-shore

Page 13: How Well Forecast Were the 2004 and 2005 Atlantic and U.S. Hurricane Seasons? Mark Saunders and Adam Lea Department of Space and Climate Physics Benfield

Slide 13RMetSoc Hurricanes March 2006

3.3. Seasonal Forecast ComparisonSeasonal Forecast Comparison

Page 14: How Well Forecast Were the 2004 and 2005 Atlantic and U.S. Hurricane Seasons? Mark Saunders and Adam Lea Department of Space and Climate Physics Benfield

RMetSoc Hurricanes March 2006 Slide 14

Organisations Providing Organisations Providing Seasonal ForecastsSeasonal Forecasts

Seasonal Atlantic hurricane forecasts for the Seasonal Atlantic hurricane forecasts for the 2004 and 2005 seasons were issued by:2004 and 2005 seasons were issued by:

Tropical Storm Risk (TSR)Tropical Storm Risk (TSR) (9 for each season) (9 for each season)

Gray/Colorado State UniversityGray/Colorado State University (6 for each season) (6 for each season)

NOAANOAA (2 for each season) (2 for each season)

Meteorological Institute, CubaMeteorological Institute, Cuba (2 for each season) (2 for each season)

Page 15: How Well Forecast Were the 2004 and 2005 Atlantic and U.S. Hurricane Seasons? Mark Saunders and Adam Lea Department of Space and Climate Physics Benfield

RMetSoc Hurricanes March 2006 Slide 15

U.S. ACE IndexU.S. ACE Index

• Combines storm numbers, intensity and duration Combines storm numbers, intensity and duration within a single index within a single index (Saunders and Lea, 2005).(Saunders and Lea, 2005).

• Defined as the sum ofDefined as the sum of the squares of hourly the squares of hourly maximum 1-min sustained winds for all storm maximum 1-min sustained winds for all storm systems over the U.S. mainland while they are at systems over the U.S. mainland while they are at least tropical storm in strength. least tropical storm in strength.

• Strongly linked to U.S. hurricane Strongly linked to U.S. hurricane insured lossinsured loss (Collins and Lowe, 2001)(Collins and Lowe, 2001) with a with a rank correlation of rank correlation of 0.70.7 for the period 1900-2005. for the period 1900-2005.

The U.S. ACE (The U.S. ACE (Accumulated Cyclone EnergyAccumulated Cyclone Energy) Index:) Index:

Page 16: How Well Forecast Were the 2004 and 2005 Atlantic and U.S. Hurricane Seasons? Mark Saunders and Adam Lea Department of Space and Climate Physics Benfield

Slide 16RMetSoc Hurricanes March 2006

Deterministic Forecasts for 2005Deterministic Forecasts for 2005

Page 17: How Well Forecast Were the 2004 and 2005 Atlantic and U.S. Hurricane Seasons? Mark Saunders and Adam Lea Department of Space and Climate Physics Benfield

Slide 17RMetSoc Hurricanes March 2006

Deterministic Forecasts for 2005Deterministic Forecasts for 2005

5.55.5

Page 18: How Well Forecast Were the 2004 and 2005 Atlantic and U.S. Hurricane Seasons? Mark Saunders and Adam Lea Department of Space and Climate Physics Benfield

Slide 18RMetSoc Hurricanes March 2006

Probabilistic Forecasts for 2005Probabilistic Forecasts for 2005

North Atlantic ACE IndexNorth Atlantic ACE Index

Page 19: How Well Forecast Were the 2004 and 2005 Atlantic and U.S. Hurricane Seasons? Mark Saunders and Adam Lea Department of Space and Climate Physics Benfield

Slide 19RMetSoc Hurricanes March 2006

Probabilistic Forecasts for 2005Probabilistic Forecasts for 2005

U.S. ACE IndexU.S. ACE Index

Page 20: How Well Forecast Were the 2004 and 2005 Atlantic and U.S. Hurricane Seasons? Mark Saunders and Adam Lea Department of Space and Climate Physics Benfield

Slide 20RMetSoc Hurricanes March 2006

Probabilistic Forecasts for 2004Probabilistic Forecasts for 2004

North Atlantic ACE IndexNorth Atlantic ACE Index

Page 21: How Well Forecast Were the 2004 and 2005 Atlantic and U.S. Hurricane Seasons? Mark Saunders and Adam Lea Department of Space and Climate Physics Benfield

Slide 21RMetSoc Hurricanes March 2006

Probabilistic Forecasts for 2004Probabilistic Forecasts for 2004

U.S. ACE IndexU.S. ACE Index

Page 22: How Well Forecast Were the 2004 and 2005 Atlantic and U.S. Hurricane Seasons? Mark Saunders and Adam Lea Department of Space and Climate Physics Benfield

Slide 22RMetSoc Hurricanes March 2006

U.S. Hurricane Prediction ModelU.S. Hurricane Prediction Model

Saunders, M. A. and A. S. Lea, Seasonal prediction of Saunders, M. A. and A. S. Lea, Seasonal prediction of hurricane activity reaching the coast of the United States, hurricane activity reaching the coast of the United States, NatureNature, 434, 1005-1008, 2005., 434, 1005-1008, 2005.

• First example of useful skill for First example of useful skill for predicting seasonal US landfalling predicting seasonal US landfalling hurricane activity and damage.hurricane activity and damage.

• The model has a The model has a sound physical basissound physical basis. . • The model will benefit risk awareness The model will benefit risk awareness

and offers good potential for and offers good potential for application in business decision application in business decision makingmaking. .

Page 23: How Well Forecast Were the 2004 and 2005 Atlantic and U.S. Hurricane Seasons? Mark Saunders and Adam Lea Department of Space and Climate Physics Benfield

Slide 23RMetSoc Hurricanes March 2006

4. Forecast Business 4. Forecast Business ApplicationApplication

Page 24: How Well Forecast Were the 2004 and 2005 Atlantic and U.S. Hurricane Seasons? Mark Saunders and Adam Lea Department of Space and Climate Physics Benfield

RMetSoc Hurricanes March 2006 Slide 24

BackgroundBackground

• The TSR early August forecast model The TSR early August forecast model (Saunders (Saunders and Lea, 2005)and Lea, 2005) correctly anticipates whether U.S. correctly anticipates whether U.S. hurricane losses are above-median or below-hurricane losses are above-median or below-median in median in 74% of years74% of years between 1950 and 2003. between 1950 and 2003.

• This skill combined with the success of the This skill combined with the success of the seasonal U.S. landfalling hurricane forecasts for seasonal U.S. landfalling hurricane forecasts for 2004 and 2005, suggests that forecast precision 2004 and 2005, suggests that forecast precision may now be may now be high enough to offer potential high enough to offer potential benefit to industriesbenefit to industries whose returns are affected whose returns are affected by hurricane damage. by hurricane damage.

Page 25: How Well Forecast Were the 2004 and 2005 Atlantic and U.S. Hurricane Seasons? Mark Saunders and Adam Lea Department of Space and Climate Physics Benfield

Slide 25RMetSoc Hurricanes March 2006

Hurricane Loss ProbabilityHurricane Loss ProbabilityUS hurricane US hurricane total insured total insured loss contingent loss contingent on the TSR on the TSR (Tropical Storm (Tropical Storm Risk) 1st August Risk) 1st August forecast.forecast.

The chance of a The chance of a large total loss large total loss is much higher is much higher in those years in those years when the when the forecast is high.forecast is high.

Page 26: How Well Forecast Were the 2004 and 2005 Atlantic and U.S. Hurricane Seasons? Mark Saunders and Adam Lea Department of Space and Climate Physics Benfield

RMetSoc Hurricanes March 2006 Slide 26

2004 and 20052004 and 2005

• For the damaging 2004 and 2005 hurricane For the damaging 2004 and 2005 hurricane seasons the TSR early August forecasts seasons the TSR early August forecasts predicted U.S. landfalling hurricane activity predicted U.S. landfalling hurricane activity in the in the upper quartileupper quartile and and upper decileupper decile respectively.respectively.

• Thus TSR would have recommended that Thus TSR would have recommended that insurance companies insurance companies purchase extra purchase extra protection.protection.

Page 27: How Well Forecast Were the 2004 and 2005 Atlantic and U.S. Hurricane Seasons? Mark Saunders and Adam Lea Department of Space and Climate Physics Benfield

RMetSoc Hurricanes March 2006 Slide 27

5. Conclusions5. Conclusions

• 2004 and 2005 Atlantic and U.S. landfalling 2004 and 2005 Atlantic and U.S. landfalling hurricane seasons were hurricane seasons were both predicted to be both predicted to be activeactive (upper tercile activity to high probability) (upper tercile activity to high probability) from the previous December.from the previous December.

• Overall the Overall the TSR forecastsTSR forecasts slightly outperformed slightly outperformed those from the other forecast groups (certainly those from the other forecast groups (certainly the case for U.S. landfalling activity). the case for U.S. landfalling activity).

• Seasonal forecast precision now appears high Seasonal forecast precision now appears high enough to be enough to be practically useful.practically useful.