forecasting monthly tropical cyclone activities
DESCRIPTION
Forecasting Monthly Tropical Cyclone Activities. Lee Sai Ming Hong Kong Observatory, Hong Kong, China ESCAP/WMO Integrated Workshop on Urban Flood Risk Management in a Changing Climate: Sustainable and Adaptation Challenges 6 Sep 2010, Macao, China. The need for long-range TC forecast. - PowerPoint PPT PresentationTRANSCRIPT
1
Forecasting Monthly Tropical Cyclone Activities
Lee Sai MingHong Kong Observatory, Hong Kong, China
ESCAP/WMO Integrated Workshop on Urban Flood Risk Management in a Changing Climate:
Sustainable and Adaptation Challenges6 Sep 2010, Macao, China
The need for long-range TC forecast
2
To support decision-makingTo support disaster prevention and
preparedness planning
Availability of seasonal TC forecast
3
WMO Bulletin 56 (4) : Seasonal Tropical Cyclone Forecasts – a very comprehensive overview
Examples of forecasts:No. of TC / named storms / ACE index over an
ocean basin [no region-specific info.]
No. of TC landfalls [TC affecting a region/city does not need to make landfall there]
Typhoon Chanthu, July 2010
4
Typhoon Hagupit, September 2008
5
Heavy rain brought by Typhoon Chanthu
6
>= 100 mm of rainfall
(courtesy of TVB)
Flooding in Tai O after Typhoon Hagupit
TC activity affecting a region / city
8
Definition: No. of TC within a certain range and a certain period of time
Hong Kong: N500 [within 500 km of HK]
Long term mean of annual N500 ≈ long term mean of annual Nsig [issuance of warning signals]
Monthly N500 of Hong Kong
9
HK TC season ~ June to October
July N500 of HK
10
Forecast formulation
11
N500 is a count parameterCan be modelled by the Poisson distribution
The Poisson dist. belongs to the family of exponential dist.
Canonical form:
!),(
y
eyp
y
,...2,1,0y
)]()()()(exp[);( ydcbyayf
where a(y)=y, b(θ)=log θ, c(θ)= θ, d(y)=-logy!
Generalized Linear Model (GLM)
12
Tii xYEg ))((
Yi = response variable [monthly N500]
E = expectation of the dist.
g = link function
xi = explanatory variables or predictors
β = model parameters
i = 1, 2, … N
What are the predictors ?
NCEP CFS
13
Climate Prediction Center, NOAA, USA: a WMO designated Global Producing Centre (GPC) of Long Range Forecasts
CPC provides digital long range forecast and hindcast [generated by the NCEP Climate Forecast System]
Hindcast used in this study: 1981-2008Variables: mslp, upper air wind, stream
function, geopotential height, SST, vorticity, divergence … [26 variables]
14
Spatial coverage of data used
Atmospheric variables and SST
10S – 50N, 90E – 150W
Eq. Pacific SST:15S – 15N, 150E – 80W
Data compression
15
Horizontal resolution of data: 1 lat. x 1 lon. for SST 2.5 lat. x 2.5 lon. for others
No. of data grid points >= 1225 [for each variable]
Compress data by EOF analysis1225 data points 28 principal components
Selection of predictors and combinations
16
1. Fit a single predictor GLM, search for skilful single predictor
2. Fit a multiple predictor GLM [predictors from step
1], filter out redundant predictors by stepwise regression [no. of combinations >= 2 x 108, hence randomly select a limited no. of combinations]
Cross-validate the ‘reduced GLM’ [from step 2], search for top performers [a ‘brute force’ approach]
Cross-validation
17
1. Hide the observation of 1 year
2. Estimate the GLM parameter from the rest of the observations and the principal components
3. Verify the GLM forecast against the hidden observation
4. Rotate the process through 28 years
Cross-validation result provides skill estimates for real-time forecasts.
A low-cost / no-cost tool
18
Performance comparison against the climatological forecastHindcat period: 1981-2008
19
Climatology(mode)
1971-2000Top GLM Gain (%)
Jun 14 24 71
Jul 16 23 44
Aug 15 22 47
Sep 14 20 43
Oct 16 25 56
Hindcast Vs Actual July N500 (1981-2008)
20
Physical Interpretation
21
The 4th EOF of 500 hPa geopotential height of June
Physical Interpretation
22
The 1st EOF of 850 hPa zonal wind of October
Multi-GLM combination
23
Weigel et al., 2008: Can Multi-model Combination Really Enhance the Prediction Skill of Probabilistic Ensemble Forecasts? Quarterly Journal of the Royal Meteorological Society
A message in respect of deterministic forecasts: combination of similarly skilful models can enhance prediction skill
Further exploitation of the cross-validation result
Performance comparison against the climatological forecastHindcat period: 1981-2008
24
Climatology(mode)
1971-2000Top GLM
Multi-GLM (mode of top
20 GLM)
Jun 14 24 27
Jul 16 23 25
Aug 15 22 26
Sep 14 20 23
Oct 16 25 28
Conclusion
25
Monthly TC forecast can be formulated in terms of GLM
Dynamical climate model (e.g. NCEP CFS) forecast data contain a lot of predictive information
The ‘brute force’ approach is viable in identifying skilful predictors and combinations
Further skill enhancement is made possible by multi-GLM combination
Remarks
26
Too many predictors: inexhaustible combinations.
Not all EOF can be easily interpreted.
27
Thank you
Acknowledgement:
The Hong Kong Observatory gratefully acknowledges NOAA/CPC for providing CFS forecast and hindcast data on the web to support research and seasonal forecasting operation conducted by the Observatory.
Table 6a. Top 20 GLM combinations for June. The number k after the underscore indicates the kth PC. Same convention for Table 6b to 6e. Meaning of the variable is given in Table 1.
div850_21 pwat_11 sf200_1 z500_4
mslp_23 vp850_8 z200_19 z500_4
sf200_7 v200850_28 z200700_21 z200_19
tmp2m_7 z200_19 z200_3 z500850_13
csstf_7 v850_16 v850_21
csstf_7 pwat_11 v850_21
pwat_11 u200_28 v200850_4
mslp_2 z200_19 z500850_13
sf200_7 tmp2m_7 v200850_18 v850_21
pacsst_24 sf200_1 z500_4
vp850_8 z200_23 z200_3 z500850_13
div850_28 sf200_28 vp200_7
div850_28 pwat_11 sf850_6
sf200_1 u200_28 z200_19 z500_4
div200_18 div850_21 z200_23 z200_3
tmp2m_26 z200_19 z200_3
div200_18 sf200_28 sf200_7 sf850_28
div850_21 sf850_16 v200_19 z500_4
div850_21 sf850_16 z500_4
csstf_7 u200850_14 z500_4
Table 6b. Top 20 GLM combinations for July.
pwat_11 pwat_9 u200850_16 z200700_24
div850_15 u200850_16 z700_25
div200_17 u850_24 v200850_4
u850_24 vp850_12
div850_15 z500850_8 z700_25
div850_15 pwat_9 v200_6 vp200_19
prate_17 u200850_16 v200850_4 z700_25
u200_25 z200850_17 z500850_8
prate_17 u200850_16 v200_24 vp200_8
pwat_11 pwat_9 z200_22
div200_17 vp200_8 z700_25
prate_17 v200850_4 z200850_24
z200700_17 z700_25
div850_15 z500_24
div200_17 pwat_9 z200700_24
div850_15 pwat_9 vp200_19
sf200_20 vp200_8 z200_27
u850_24 vp200_8
sf850_23 vp850_12
pwat_11 pwat_9 u200850_16 z200850_24
Table 6c. Top 20 GLM combinations for August.
u200850_17 u850_23 vor850_28
tmp2m_1 u850_23 v200_24 z200700_17
sf850_17 u850_23 vor850_28
tmp2m_1 z200700_17 z500_14
vor850_28 z200_13
u200850_17 vor850_28 z500_14
pacsst_15 prate_24 u850_12
u200850_15 u200850_17 vor850_28
tmp2m_1 vp200_26 z850_12
pacsst_4 sf850_13 vp200_26
tmp2m_1 u850_23 z200700_17
csstf_20 u850_12 vor850_28
tmp2m_1 vp200_26 z500_14
sf850_17 vor850_28 vp200_26
prate_24 u200850_17 vor850_28
pwat_26 tmp2m_1 z500_14
pwat_26 vor850_28 z500_14
sf850_17 vor850_28 vp850_18
div850_28 prate_24 vor850_28
csstf_20 sf850_17 vor850_28
Table 6d. Top 20 GLM combinations for September.
sf850_16 u200850_11 vor850_13
div200_11 sf850_16 z700_5
div200_25 z500_16
div200_11 v200850_16 vp200_24
vp850_13 z500_16
div200_25 div850_21 pwat_23
mslp_19 z500_16
prate_12 z500_16
div200_25 u200850_11 z500_16
div200_25 div850_21 vp200_5
prate_17 prate_19 vor850_24
div200_25 u200850_11 vp200_5
prate_12 prate_19 vor850_26
vp200_24 z500_16
prate_12 prate_19 u200850_11
div200_11 div200_25 vor850_24
div200_11 v200_18 z700_5
div200_25 vor850_13 vor850_24
div200_11 div200_25 u200850_11
pwat_17 vp200_24 z500_16
Table 6e. Top 20 GLM combinations for October.
div200_20 prate_10 vor850_2 z200_2
div200_11 sf200_11 z500850_4
csstf_19 div200_11 z500850_4
tmp2m_23 v200850_25 z500850_4 z500_3
csstf_19 vor850_1 z500850_4
pwat_1 sf850_1 u850_1 v850_17 z500_1
u200850_12 v200_24 z500850_4 z500_10
pwat_18 tmp2m_3 z850_3
v200850_13 v850_25 z700_1
div200_20 u850_4 z200_1 z700_3
v200850_13 v850_25 z500_1
v200850_13 v200_15 v200_24 vor850_28
sf850_6 v200850_13 v850_25
mslp_2 u850_4 vor850_4 z200850_2 z200_2
prate_10 pwat_18 u200_3
v200850_13 z700_2 z850_3
mslp_3 pwat_18 z850_2 z850_3
csstf_19 div850_21 z500850_4 z500_3
sf850_28 u200_3 z500_1
prate_21 sf850_28 vp850_18