tropical cyclone forecasts dr. richard j. murnane risk prediction initiative bermuda biological...

42
Tropical Cyclone Tropical Cyclone Forecasts Forecasts Dr. Richard J. Murnane Risk Prediction Initiative Bermuda Biological Station for Research, Inc.

Upload: clement-robbins

Post on 17-Dec-2015

217 views

Category:

Documents


3 download

TRANSCRIPT

Tropical Cyclone ForecastsTropical Cyclone Forecasts

Dr. Richard J. MurnaneRisk Prediction Initiative

Bermuda Biological Station for Research, Inc.

Hurricane Landfall in Miami Hurricane Landfall in Miami and Dade Countyand Dade County

Cat 1&2 Cat 3, 4, 5

1910 1930 1950 1970 1990

Hu

rric

an

e L

and

fall

per

Ye

ar 2

1

0

Year

Landsea, 1997

Tropical Cyclone Tropical Cyclone Seasonal Forecasting Seasonal Forecasting

StrategyStrategy Identify predictable global and Identify predictable global and

local features that drive climatelocal features that drive climate

Statistically relate features to Statistically relate features to tropical cyclone occurrencetropical cyclone occurrence

OutlineOutline Background on tropical cyclonesBackground on tropical cyclones

– Necessary conditionsNecessary conditions

– Forecast parametersForecast parameters

Seasonal forecastsSeasonal forecasts

– Basin-wideBasin-wide

– LandfallLandfall

Real-time hurricane intensity forecastsReal-time hurricane intensity forecasts

– Numerical predictionNumerical prediction

– Theoretical Maximum Potential IntensityTheoretical Maximum Potential Intensity

Necessary ConditionsNecessary Conditions Coriolis Effect

Unstable Atmosphere

Sea Surface Temperature > 26º C

Trigger

Little Vertical Wind Shear

High Relative Humidity

Major Climatic Features Affecting Major Climatic Features Affecting Atlantic Basin Hurricane ActivityAtlantic Basin Hurricane Activity

El NiEl Niño-Southern Oscillation (ENSO)

Quasi-biennial Oscillation

Sea Surface Temperature (SST)

West Africa Rainfall

Atmospheric Zonal Wind Anomalies

Sea Level Atmospheric Pressure

El Niño-Southern OscillationEl Niño-Southern Oscillation

Images from IRI web site

Normal

El Niño

US Hurricane Landfalls US Hurricane Landfalls Cat. Cat. 33

0.0

0.2

0.4

0.6

0.8

2 4 6 80

Pro

babi

lity

Number Landfalls Per Year

El Niño

Neutral

El Viejo

J. O’Brien & COAPS

0.0

0.2

0.4

0.6

0.8

1.0

Cat 1 Cat 2 Cat 3

El Niño

Neutral

El Viejo

Baroclinically InfluencedBaroclinically InfluencedAtlantic Storms:Atlantic Storms:

Hurricane Landfalls/yearHurricane Landfalls/year

# La

ndfa

lls/y

ear

Tropical Only Atlantic Storms:Tropical Only Atlantic Storms:Hurricane Landfalls/yearHurricane Landfalls/year

# La

ndfa

lls/y

ear

0.0

0.5

1.0

1.5

Cat 1 Cat 2 Cat 3

El Niño

Neutral

El Viejo

Quasi-Biennial Oscillation Quasi-Biennial Oscillation (QBO)(QBO)

Arrows show stratospheric wind direction

West Phase

Enhanced Atlantic Hurricane Formation

East Phase

Suppressed Atlantic Hurricane Formation

Sea Surface TemperatureSea Surface Temperature

Climate Prediction Center

Sea Surface Temperature AnomaliesSea Surface Temperature Anomalies

Climate Prediction Center

Seasonal Atlantic Hurricane Activity (1886-1991)

0

100

200May

June

July

Aug

Sept

Oct

Nov

Dec

Jan-Apr

West African RainfallWest African Rainfall

WesternSahel Region

Gulf of GuineaRegion

-2.00

-1.50

-1.00

-0.50

0.00

0.50

1.00

1.50

2.00

Sahel Rain (1899 – 1990)

1900 1920 1940 1960 1980

Sta

nd

ard

De

via

tion

Landsea et al., 1992

Year

2

1

0

-1

-2

1949 – 19901949 – 1990

0.64 IHD

7 Driest Sahel Years7 Driest Sahel Years 7 Wettest Sahel Years7 Wettest Sahel Years

14.43 IHD

Gray & Landsea, 1992

Intense Hurricanes vs. Sahel RainIntense Hurricanes vs. Sahel Rain

Long-term TrendsLong-term Trends

Year

Sta

nd

ard

Dev

iati

on

N. Atlantic SST AnomalyNE Atlantic SLP Anomaly (neg)Sahel RainIntense Hurricane Days

1950 1960 1970 1980 1990

2

1

0

-1

-2

Gray, 1998

Basin-Wide Seasonal Basin-Wide Seasonal ForecastsForecasts

Gray et al.’s Basin-wide ForecastsGray et al.’s Basin-wide Forecasts

Issued in: December, April, June, AugustIssued in: December, April, June, August

Forecast includes a variety of parameters:Forecast includes a variety of parameters:– Named StormsNamed Storms– HurricanesHurricanes, Hurricane Days, Hurricane Days– Intense HurricanesIntense Hurricanes, Intense Hurricane Days, Intense Hurricane Days– Hurricane Destruction PotentialHurricane Destruction Potential– Net Tropical Cyclone ActivityNet Tropical Cyclone Activity

Predictive Parameters Vary With ForecastPredictive Parameters Vary With Forecast

Gray et al.’s ‘98 Atlantic Gray et al.’s ‘98 Atlantic Basin ForecastsBasin Forecasts

1950-90 Average Dec April June Aug Total as of1 Oct 1998

Named Storms (9.3) 9 10 10 10 11

Hurricanes (5.8) 5 5 6 6 7

Intense Hurricanes (2.2) 2 2 2 2 2

Net Tropical CycloneActivity (100%)

90% 95% 100% 110% 121%

Landsea, 1997

Landsea, 1997

Elsner’s Atlantic Basin Elsner’s Atlantic Basin ForecastsForecasts

Uses subset of Gray variablesUses subset of Gray variables– Zonal windsZonal winds– West African rainfallWest African rainfall

Divides storms into tropical only and Divides storms into tropical only and baroclinically influencedbaroclinically influenced– assumes climatology for baroclinically assumes climatology for baroclinically

influenced storms (influenced storms (per year)per year)– Predicts tropical only storms using OLSPredicts tropical only storms using OLS

Calculates intense hurricanes using Calculates intense hurricanes using Poisson regressionPoisson regression

Elsner et al. Forecast 1998 Elsner et al. Forecast 1998 Atlantic Intense HurricanesAtlantic Intense Hurricanes

Elsner et al., 1998

0

0.1

0.2

0.3

0.4

0.5

1 2 3 4 5 6 7

Number of Storms

Pro

bab

ilit

y

0 1 2 3 4 5 6

August ‘98 ForecastsAugust ‘98 Forecasts

Mean1950-1990

Gray Elsner Observed8 Aug 98

Observed1 Oct 98

Named Storms

Hurricanes

9.3

5.8

10

6

-

3

5

3

11

7

Intense Hurricanes 2.3 2 1 1 2

Net TropicalCyclone Activity

100% 110% - 59% 121%

Seasonal Landfall and Seasonal Landfall and Location ForecastsLocation Forecasts

Gray Landfall Forecast Gray Landfall Forecast

Divides coast into separate sections Divides coast into separate sections

Uses Atlantic Basin seasonal forecast Uses Atlantic Basin seasonal forecast of “Net Tropical Cyclone Activity” (NTC) of “Net Tropical Cyclone Activity” (NTC) and SST anomalies to calculate total and SST anomalies to calculate total landfall probabilitylandfall probability

Distributes total probability along coast Distributes total probability along coast based on landfall climatologybased on landfall climatology

Cat 1&2 Hurricane LandfallsCat 1&2 Hurricane Landfalls

Gray, 1998

Intense Hurricane LandfallsIntense Hurricane Landfalls

Gray, 1998

1998 Landfall % Probability 1998 Landfall % Probability ForecastForecast

368

39

10.2 34

33

4

7

5

6 8

Gray, 1998

Elsner Landfall Elsner Landfall Probability ForecastsProbability Forecasts

Uses “Logistic Regression” for predictionsUses “Logistic Regression” for predictions

Predictors include:Predictors include:– Zonal wind anomaliesZonal wind anomalies– Sea level pressuresSea level pressures– QBOQBO– West African rainfallWest African rainfall

1998 Location Forecast1998 Location Forecast

IntenseHurricane in

the Gulf

IntenseHurricane in

the Caribbean

Hurricane Landfallon Southeast US

coast

Probability 0.189 0.107 0.469

Elsner et al., 1998

Hurricane Intensity Hurricane Intensity ForecastsForecasts

NHC Intensity Forecasts (1990-97)

25

20

15

10

5

Err

or

(kn

ots

)

1990 1991 1992 1993 1994 1995 1996 1997

72 hours

48 hours

24 hours

Avila, 1998

Year

Ginis et al.’s Hurricane Ginis et al.’s Hurricane ForecastsForecasts

Based on Coupled Ocean-Atmosphere Based on Coupled Ocean-Atmosphere ModelModel

Full Ocean Model Reproduces Air-Sea Full Ocean Model Reproduces Air-Sea Interactions and Improves Intensity Interactions and Improves Intensity ForecastsForecasts

Hurricane OpalHurricane Opal

Ginis, 1998

Maximum Potential Intensity Maximum Potential Intensity (MPI)(MPI)

Developed by Emanuel and BisterDeveloped by Emanuel and Bister

MPI Controlled by Environmental MPI Controlled by Environmental Factors, e.g.:Factors, e.g.:– Sea Surface TemperatureSea Surface Temperature

– Upper Atmosphere TemperatureUpper Atmosphere Temperature

– Relative HumidityRelative Humidity

Doesn’t account for forward motionDoesn’t account for forward motion

Contours of Maximum Contours of Maximum Wind Speed (m/s)Wind Speed (m/s)

Emanuel and Bister, 1998

Sea Surface Temperature (ºC)

Up

pe

r A

tmo

sp

he

re

Te

mp

era

ture

(ºC

)

15 20 25 30 35 40 45

-100

-75

-50

-25

Maximum Potential Maximum Potential Winds (m/s)Winds (m/s)

Emanuel and Bister, 1998

Maximum Potential Maximum Potential Wind (knots)Wind (knots)

Emanuel and Bister, 1998

ConclusionsConclusions Many of the seasonal forecast parameters Many of the seasonal forecast parameters

are related to the necessary conditions for are related to the necessary conditions for tropical cyclone formationtropical cyclone formation

Basin-wide seasonal forecasts are easier to Basin-wide seasonal forecasts are easier to make than landfall forecasts but they are of make than landfall forecasts but they are of less relevanceless relevance

Realistic representations of the ocean are Realistic representations of the ocean are needed to improve tropical cyclone intensity needed to improve tropical cyclone intensity forecastsforecasts