High Resolution Ensemble Prediction of

Typhoon Morakot (2009)

Ying-Hwa Kuo1,* and Xingqin Fang1,2

1National Center for Atmospheric Research, Boulder, Colorado, USA

2Department of Atmospheric Sciences, School of Environmental Sciences

and Engineering, Sun Yat-Sen University, Guangzhou, China

May 11, 2011

From August 6 to 10, 2009, extraordinary rainfall was brought over

Taiwan by Typhoon Morakot, breaking 50 year’s precipitation record,

causing a loss of more than 700 people and estimated property damage

exceeding US$3.3 billion

Observed Rainfall of Typhoon Morakot (2009)

Typhoon Morakot (2009)

Max. 24-h gauge 1504 mm

Max. 96-h gauge 2874mm

at Chiayi County

(windward slope of CMR)

Accumulated rainfall:

(a) 96-h on August 6-10

(b) 24-h on August 8-9

* Objective analysis

~450 automatic stations

24-h rain world record

1825 mm

Data and Methodology --- Experiment Design

• Two deterministic simulations with and without Taiwan topography: ec0600 and ecNT0600,

ICBC from ECMWF high resolution analysis (0.225º×0.225º )

• Two sets of ensemble simulations with and without Taiwan topography: EN0600 (m1-32)

and NTEN0600 (m1-32), ICBC perturbation by WRF 3DVAR

• 144h simulation started from 0000UTC 6 to 0000 UTC 12 Aug. 2009

• Domains: 2-way 36-km (280×172), 12-km (430×301) and 4-km (364×322)

• 36 levels to the top of 20 hPa

• ARW WRF physics: WSM5, YSU, Noah, RRTM, Goddard, BMJ (all)

Taiwan topography is

removed for sensitive

experiment, but its land

surface features still

retained.

Question: What is the role of Taiwan topography in this extreme rainfall event?

With Taiwan topography• Intensive rainfall areas (>800 mm) well captured• Extremes (>2500 mm) captured, with displacements• Peak 3128 mm

Without Taiwan topography• Even rainfall distribution• No obvious local rainfall enhancement • Peak 616 mm, less than 20%

Taiwan topography enhances and focuses the rainfall along the windward side of the mountain and thus

greatly amplifies the local rainfall extremes in the spatial distribution of rainfall, it plays a key role in making

Typhoon Morakot a record-breaking rainfall event.

96-h Rain

EN0600 mean

96-h Rain

NTEN0600 meanOBS

OAR

* OAR: Orographically Additive Rainfall

OAR= EN0600_mean - NTEN0600_mean

• 3-h OAR is positive during the entire simulation

• 3-h OAR is up to about 25 mm after landfall

• 96-h OAR is 382 mm

3-h rainfall rate from ensemble system

with and without topography

landfall left Taiwan

The simulated track of ensemble members (green line), the single deterministic

simulation (blue line), and the average track of the ensembles (red line) of EN0600

(top) and NTEN060 (bottom). The JMA best track (modified by analysis from Taiwan

Central Weather Bureau) is superimposed by thick black line as OBS.

The variability of the storm track and intensity

The rainfall probability distribution (%) exceeding the thresholds of (a) 500, (b) 1000 mm for 24-h

rainfall ending at 0000 UTC 8 August; (c) 500, (d) 1000 mm for 24-h rainfall ending at 0000 UTC 9

August; (e) 1000, (f) 1500 mm for 48-h rainfall ending at 0000 UTC 9 August; (g) 1500, (h) 2500

mm for 96-h rainfall ending at 0000 UTC 10 August estimated from the 32 members of EN0600. The

observed rainfall at the corresponding threshold is superimposed by the blue line.

a. The rainfall probability forecast

(a) D2

500 mm D200/7-00/8

(b) D2

1000 mm

(c) D3

500 mm

(d) D3

1000 mm

(e) 2Ds

1000 mm

(f) 2Ds

1500 mm

(g) 4Ds

1500 mm

(h) 4Ds

2500 mm

D300/8-00/9

4Ds00/6-00/10

2Ds00/7-00/9

Data and Methodology --- Experiment Design

• Six sets of ensemble experiments with different CU settings in the 3 2-

way domains, 36-km (280×172), 12-km (430×301) and 4-km (364×322):

CU_01 BBB CU_03 BBE CU_05 BEE

CU_07 KKK CU_09 KKE CU_11 KEE

• Each ensemble set has the same 8-member ensemble ICBC:

perturbed from ECMWF high resolution analysis (0.225º×0.225º ) by WRF

3DVAR

• 96h simulation started from 0000UTC 6 to 0000 UTC 10 Aug. 2009

• 36 levels to the top of 20 hPa

• ARW WRF physics: WSM5, YSU, Noah, RRTM, Goddard

Taiwan topography is

removed for sensitive

experiment, but its land

surface features still

retained.

Ensemble average

track

01 BBB 03 BBE 05 BEE

07 KKK 09 KKE 11 KEE

13 NC

Single deterministic

simualtion track

No Terrain

With Terrain

FNL

EC

SLP 500 hPa Z

ICFNL_36km

ICEC_12km

Summary

• Simulations with BMJ scheme on 36-km

and 12-km all have westward and

southward track bias.

• Simulations with Kain-Fritsch scheme do

not have such bias

• Without topography, the storm tracks tend

to shift northward:

– Without CMR, southwesterly monsoon flow

can push the storm track further north

Ensemble

mean

96-h rainfall

00/6-00/10

OBS

Four-day accumulated ensemble mean rainfall from the

various cumulus parameterization experiments

KKK

KKE

KEE

OBS

BBB

BBE

BEE

ALL

2d-cape and 850 hPa wind vector

(M4_BBB, M4_BBE, and M4_BEE)

2d-cape and 850 hPa wind vector

(M4_KKK, M4_KKE, and M4_KEE)

M4_BBB

M4_BBE

M4_BEE M4_KEE

M4_KKE

M4_KKK

BBE(CU_03) and BEE(CU_05)

CU sensitivity:

BBE(&BBB) and BEE differ much

KKE(&KKK) and KEE do not differ much

BEE and KEE do not differ much

KKE(CU_09) and KEE(CU_11)

Implication:

--- Using BMJ on 12 km resolution may have important impacts on low level

and high level flow patterns; KF does not have these impacts.

--- If CU is used only on 36 km resolution, the tracks are not sensitive to CU.

Diagnostics on CU sensitivity

--- M4 with BBE and BEE

M4_BBE and M4_BEE share similar westbound tracks from 00/6 to

12/7, they separate around 00/8, about 50 km away from Taiwan.

650 hPa smoothed absolute

vorticity and wind (M4_BBE)

650 hPa smoothed absolute

vorticity and wind (M4_BEE)

SLP and surface wind

(M4_BBE)

SLP and surface wind

(M4_BEE)

Analysis of 53 typhoons that came close to Taiwan, 1946-1975 by Wang

(1980).

Strong Typhoons Weak Typhoons

Does the super ensemble perform better than any

sub-ensemble? Do you find out any best CU?

• Track --- Yes

• Rainfall --- Yes, improve rainfall over Chiayi and southern Taiwan

• Best CU --- Not sure at this point. Only test two Cu Pa schemes, also

only one case

Summary• Deterministic high-resolution prediction of the extreme

rainfall event of Morakot (2009) is exceedingly difficult,

as uncertainties in initial conditions and model physics

can have significant influence on storm tracks and

rainfall prediction.

• Probability forecast using high-resolution ensemble can

provide useful information on extreme rainfall associated

with Morakot (2009).

• The performance of high-resolution ensemble is

sensitive to model physics:

– Use of BMJ scheme on 12-km grid produced a weaker storm

track with westward and southward bias; while Kain-Fristch

scheme did not.

– Super ensemble, including both uncertainties in initial conditions

and physics, gave superior performance in track forecasts