1
Subseasonal Variability of Hurricane Activity
Kathy Pegion
Center for Ocean-Land-Atmosphere Studies
Philip Pegion (CPC),
Tim DelSole (COLA), & Mihai Sirbu (COLA)
Tropical cyclones tend to cluster in time and space on subseasonal timescales.
Is this predictable?
Can it be forecasted with a dynamical model?
Counts of Atlantic TS and Hurricanes every 6 hrs
Courtesy of the Clivar MJOWG 3
Week 2 Outlook – Valid: Sep 23 – 29, 2008 Issued: 9/15
1. An increased chance for tropical cyclogenesis for the eastern Pacific, western Gulf Of Mexico and western Caribbean. The favorable phase of the MJO increases the threat for tropical development during the period. Confidence: Moderate
2. An increased chance for above-average rainfall for the eastern Pacific, southern Mexico and parts of Central America. The enhanced phase of the MJO is expected to result in wet conditions during the period. Confidence: Moderate
3. An increased chance for below-average rainfall for the equatorial Indian Ocean and western Indonesia. The suppressed phase of the MJO is expected to result in dry conditions during the period. Confidence: Moderate
4. An increased chance for above-average rainfall stretching from Southeast Asia into the western Pacific. The enhanced phase of the MJO is expected to result in wet conditions during the period. Confidence: Moderate
5. An increased chance for tropical cyclogenesis for the South China Sea and the far western Pacific. The favorable phase of the MJO increases the threat for tropical development during the period. Confidence: Moderate
SEE TEXT NOTATION: Conditions are expected to begin to become more favorable for tropical cyclogenesis across the deep tropical Atlantic Ocean late during the period but development is more likely during week 3.
5
1. Can the relationship between MJO and tropical cyclone activity be captured using common hurricane indices (eg. SGP and GPI)?
2. How well does the CFS simulate the relationship?
3. Can forecasts of these indices by the CFS be used to predict where tropical cyclone formation is more/less likely on 1-4 week timescales?
Questions
Seasonal Genesis Parameter (Gray 1979)
SGP=vorticity*coriolis*shear*thermal energy*moist stability *RH
Genesis Potential Index (Emanuel and Nolan 2004) GPI=absolute vorticity*RH * Potential Intensity * shear
Hurricane Indices
7
Data
NCEP Reanalysis-2 1979-2007
CFS03 (T62L64) 52-year simulation
Filtering
1. Five-day running mean applied to fields prior to calculating SGP or GPI
2. ENSO removed from fields prior to calculating SGP or GPI using linear regression with Nino 3.4
All tropical storm formations from the Hurricane Best Track Database 1979-2007
Climatology and Standard Deviation of SGP & GPI
NCEP Reanalysis-2
RMM1 9
RMM2
Phase 1
Phase 2 Phase 3
Phase 4 Phase 5 Phase 6
Phase 7
Phase 8
Maritime Continent
Indian Ocean Western Pacific
W. Hemisphere & Africa
MJO Index
SQRT(RMM12+RMM22) > 1
Real-time Multivariate MJO Index (Wheeler and Hendon 2004)
(obtained from Matt Wheeler, BMRC; 1975-2000)
1. Based on combined EOFs of U200, U850, and OLR
2. First two PC timeseries (RMM1 & RMM2) represent the MJO amplitude and phase
Composite OLR (W/m2) anomalies (Jun-Nov)
• Large-scale region of
enhanced/suppressed convection
• Eastward propagation
• Note enhanced/supressed
convection in Central America &
Atlantic
Observed CFS03
11
Composite anomalies (Jun-Nov) Based on observed RMM
•SGP & GPI vary with phase of MJO
•SGP & GPI are very similar
R2 SGP R2 GPI
Obs OLR
Composite SGP & GPI anomalies (Jun-Nov) Reanalysis-2 (1979-2007)
Location of tropical storm formations
from Hurricane Best Track Database 1979-2007 obtained from Unisys
• Actual formations are generally
consistent with regions of +/- SGP/GPI anomalies
SGP GPI
13
Quantifying the Relationship
H0: P(Hur Index | MJO Phase) = P(Hur Index)
Hurricane Index: GPI or SGP Terciles MJO Phase : 8 phases + null phase
M P
HurIndex,MJOHurIndex
3,MJO 9log P
HurIndex,MJOP
HurIndexP
MJOMutual Information
0 5 10 15 20 25 30 35
1 2 3 4 5 6 7 8 9
all year Jun-Nov
P(MJO phase)
15
GPI SGP
33% (Jun-Nov) 33% (Jun-Nov)
67% (Jun-Nov) 67% (Jun-Nov)
P(Hur Index) Each Tercile = 33%
15
P(GPI, MJO Phase)
GPI <= 33%ile
33 %ile >
GPI < 66%ile GPI >= 66%ile
Phase 1 Phase 2 Phase 3
Phase 4 Phase 5
Phase 6 Phase 7 Phase 8
17
Mutual Information between Hurricane Index and MJO
GPI SGP
17
NCEP/R2 CFS
SGP Climatology
(Jun-Nov)
SGP Standard Deviation
19
Composite SGP (Jun-Nov)
CFS is generally able to capture the overall behavior of the
relationship between the MJO and SGP
NCEP/R2 CFS03/T62
NCEP/R2 CFS03/T62 ERA40
MJO Phase 21
No/Weak MJO
0 5 10 15 20 25 30 35 40 45
1 2 3 4 5 6 7 8 9
R2:all year R2:Jun-Nov CFS:Jun-Nov
P(MJO phase)
33% (Jun-Nov)
67% (Jun-Nov)
CFS SGP
33% (Jun-Nov)
67% (Jun-Nov)
P(Hur Index) Each Tercile = 33%
R2 SGP
23
P(Hur Index) Each Tercile = 33%
CFS GPI
33% (Jun-Nov)
67% (Jun-Nov)
33% (Jun-Nov)
67% (Jun-Nov)
R2 GPI
23
CFS SGP
Mutual Information between SGP and MJO
R2 SGP
25
Mutual Information between GPI and MJO
R2 GPI CFS GPI
25
Conclusions
1. SGP & GPI indices capture the relationship between MJO and tropical cyclone development.
2. The CFS is generally able to capture the relationship 3. Mutual information between SGP and MJO shows a
significant relationship in the eastern Pacific.
4. Mutual information between GPI and MJO shows a significant relationship throughout the domain.
27
Week 2 Outlook – Valid: Sep 23 – 29, 2008
Please note: Confidence estimates are subjective in nature and are not based on an objective scheme. The estimates are given to provide additional information to the user.
Issued: 9/15
1. An increased chance for tropical cyclogenesis for the eastern Pacific, western Gulf Of Mexico and western Caribbean. The favorable phase of the MJO increases the threat for tropical development during the period. Confidence: Moderate
2. An increased chance for above-average rainfall for the eastern Pacific, southern Mexico and parts of Central America. The enhanced phase of the MJO is expected to result in wet conditions during the period. Confidence: Moderate
3. An increased chance for below-average rainfall for the equatorial Indian Ocean and western Indonesia. The suppressed phase of the MJO is expected to result in dry conditions during the period. Confidence: Moderate
4. An increased chance for above-average rainfall stretching from Southeast Asia into the western Pacific. The enhanced phase of the MJO is expected to result in wet conditions during the period. Confidence: Moderate
5. An increased chance for tropical cyclogenesis for the South China Sea and the far western Pacific. The favorable phase of the MJO increases the threat for tropical development during the period. Confidence: Moderate
SEE TEXT NOTATION: Conditions are expected to begin to become more favorable for tropical cyclogenesis across the deep tropical Atlantic Ocean late during the period but development is more likely during week 3.
Courtesy of CPC from the Global Tropical Benefits/Hazards Website (http://www.cpc.noaa.gov/products/precip/CWlink/ghazards/ghaz.shtml)
Caveats & Future Work
1. Indices: maybe the hurricane indices and/or the MJO index are not the right one, also test different
variables.
2. So far, we have only quantified one piece of the problem. Have not quantified P ( TC | Hur Index), P(TC | MJO).
What we really want to know is:
H0: P(TC | Hur Index & MJO) = P(TC)
3. What is the prediction skill of SGP and GPI and are
29
Week-1 Week-2
Week-3 Week-4
Correlations between CFS forecasts and NCEP/R2 (Aug-Oct 2005-2007)
Full Fields of SGP, not anomalies, no MJO filter