Cluster Analysis of Tropical Cyclone Tracks and ENSO Suzana J. Camargo, Andrew W. Robertson, International Research Institute for Climate Prediction, Columbia.

Cluster Analysis of Tropical Cyclone Tracks and ENSO

Suzana J. Camargo, Andrew W. Robertson,International Research Institute for Climate Prediction,

Columbia Earth Institute, Palisades, NY

Scott J. Gaffney and Padhraic SmythDepartment of Information and Computer Science,

University of California, Irvine, CA

Outline• Introduction• Clustering Technique• Previous works on cluster analysis and tropical cyclones• Western North Pacific Results

– Mean Regression Trajectories– Tracks– Properties of main clusters– ENSO Relationship: tracks, tracks density, NTC, ACE– Composites: SST, SLP, winds, wind shear

• North Atlantic Results– Mean Regression Trajectories and Tracks– ENSO relationship– Atlantic multi-decadal signal

• Eastern North Pacific Results– Mean Regression Trajectories and Tracks– ENSO relationship

• Summary

Introduction• Identify different track types, their seasonality and

relation with large-scale circulation and ENSO.• Importance: different track types have higher

incidence on some years and make landfall in different regions.

• New clustering technique used.• Best track datasets:

– Western North Pacific – JTWC 1950-2002.– North Atlantic – NHC 1851-2003.– Eastern North Pacific – NHC 1949-2003.

• Only tropical cyclones (TCs) with tropical storm or hurricane (typhoon) intensity (no tropical depressions).

Clustering Technique

• Developed by S.J. Gaffney and P. Smyth:- S.J. Gaffney (2004), Ph.D. thesis, University of California, Irvine.

• Mixture of polynomial regression models (curves) to fit the geographical “shape” of the trajectories.

• Extension of the standard multivariate finite mixture model to allow quadratic functions.

• Enable highly non-Gaussian density functions to be expressed as a mixture of a few PDFs.

• Fitting by maximizing the likelihood of the parameters.• Rigorous probabilistic context for clustering• Accommodate easily tropical cyclone tracks of different

lengths.

Previous works on Cluster Analysis and Tropical Cyclones

• Western North Pacific:– P.A. Harr and R.L. Elsberry, Mon. Wea. Rev. 123,

1225-1246 (1985).– J.B. Elsner and K.B. Liu, Climate Research 25, 43-54

(2003);

• North Atlantic:– J.B. Elsner, Bull. Amer. Meteor. Soc. 84, 353-356

(2003); J.B. Elsner et al., J. Climate 13, 2293-2305 (2000).

• Eastern North Pacific (TC precursors):– J.B. Mozer and J.A. Zehnder, J. Geophys. Res. – Atmos. 99, 8085-8093 (1994).

Western North Pacific Tropical Cyclones

Cluster AnalysisResults

Mean Regression Trajectories

• Appropriate number of clusters appears to be seven.

• Quantitative (out of sample likelihood) and subjective analysis.

• Two main trajectory-types: “straight-movers” and “recurvers”.

• Additional clusters: detailed differences in shape among these types.

MEAN REGRESSIONTRAJECTORIES

TRACKS

TRACKS TROPICAL CYCLONES Western North Pacific 1983-2002

Number of TCs per Cluster

Cluster A

Landfall63% Regression

Trajectory

•67% reach typhoon intensity

FIRST POSITION DENSITY

NTC ANNUAL CYCLE

TRACK DENSITY

Cluster B

Landfall61% Regression

Trajectory

50% only reach TS intensity.

FIRST POSITION DENSITY

NTC ANNUAL CYCLE

TRACK DENSITY

Cluster C

•70% reach typhoon intensity

Landfall 7% Regression

Trajectory

FIRST POSITION DENSITY

NTC ANNUAL CYCLE

TRACK DENSITY

ENSO RelationshipNTC- Number of Tropical Cyclones ACE – Accumulated Cyclone Energy

Total ACE has a well known relationshipwith ENSO (Camargo & Sobel, 2004).

Total NTC per year is not significantlycorrelated with ENSO (e.g. Wang & Chan, 2002).

Tracks El Niño years Tracks La Niña years

Cluster A

Cluster E

Cluster G

Track Density per year: Difference El Niño and La Niña years

Full basin Cluster A

Cluster GCluster E

Mean NTC and ACE per cluster and ENSOA

A

E

E G

G

SST Anomalies Composites

TCs first positions

Regression trajectory

SST and TC data for SST composites:11/81 – 12/02

Sea Level Pressure Anomalies Composites

NCEP Reanalysis and TC data for composites: 1950-2002

Anomalous Low Level Wind Composites

Wind Shear Composites

Magnitude of the total wind shear between 200hPa and 850hPa

North Atlantic Tropical Cyclones Cluster Analysis

Results

Tracks and Regression Trajectories

Tracks Atlantic named Tropical Cyclones 1970-2003.

TRACKS

Mean RegressionTrajectory

Number of TCs per cluster

ENSO Relationship

NTC correlations

ACE correlations

Tracks El Niño years Tracks La Niña years

Cluster 1

Cluster 2

Cluster 3

Named Tropical Cyclones in warm/cold ENSO years 1950-2003

Named Tropical Cyclones: 1950-2003

SST Anomalies Composites

TCs First Positions

Main Development Region

SST and TC data for SST composites: 11/81 – 12/2003

Wind shear composites

Magnitude of the total wind shear between 200hPa and 850hPa.

NCEP Reanalysis and TC data for composites: 1950-2003.

Atlantic Multi-Decadal Signal

•S.B. Goldenberg, C.W. Landsea, A.M. Mesta-Nuñez and W.M. Gray, Science 293,474-478 (2001).

Number of Major Hurricanes per cluster

SST anomalies composite

SST composites: 11/1981-12/2003

Eastern North PacificTropical CyclonesCluster Analysis

Results

Mean Regression Trajectories and Tracks

ENSO Relationship

Tracks El Niño years Tracks La Niña years

Cluster 1

Cluster 2

Cluster 3

Summary

• New clustering technique applied to Northern Hemisphere TC tracks.

• Clusters with different properties: genesis and track regions, intensity, timing.

• In all basins clusters strongly related to ENSO are identified.

• Composites of large scale fields with different characteristics for each cluster identify the factors influencing the formation and movement of TCs in each cluster.