International Journal

on Marine Navigation

and Safety of Sea Transportation

Volume 1

Number 3

September 2007

279

A Numerical Prediction System for Wind and

Sea Wave: A Typhoon Case

Jau-Ming Chen

Center of General Studies and Institute of Navigation Science and Technology,

National Kaohsiung Marine University, Kaohsiung, Taiwan

Jia-Shen Hu

Department of Shipping Technology, National Kaohsiung Marine University,

Kaohsiung, Taiwan

Wen-Laung Huang

Chinese Naval Meteorological and Oceanographic Office, Kaohsiung, Taiwan

Li-Hua Hsu

Institute of Navigation Science and Technology, National Kaohsiung Marine

University, Kaohsiung, Taiwan

ABSTRACT: A numerical model system is constructed to predict surface conditions over the open oceans for

a typhoon case. Its atmospheric and oceanic components are the Weather Research and Forecasting (WRF)

model and the NOAA WaveWatch version 3 (NWW3) model, respectively. The initial condition of the WRF

is obtained from the NCEP aviation forecast, while the WRF-predicted surface winds serve as the boundary

conditions for sea wave prediction of the NWW3.The capability of this model system is evaluated in terms of

the predictions of surface wind and sea waves associated with the typhoon Bilis (No. 0604). This typhoon

formed over the west side of Guam (141?E, 12?N) on July, 9, 2006, and moved northwestward across Taiwan

to decay over southeast China on July, 15, 2006. Its moving track is reasonably predicted by the WRF with an

averaged error of 99 km in 24-hr forecast and of 233 km in 48-hr forecast. These errors are in comparable

ranges with the official typhoon forecasts conducted by weather services in the countries around the Pacific.

The circulation pattern and intensity of surface winds and height of sea waves can be adequately portrayed by

this prediction system in advance by 48 hrs. The dangerous and navigable semicircles of the typhoon are also

clearly delineated. As such, the spatial domains of high wind and high sea are identified, providing potentially

useful information for navigation safety.

1 INTRODUCTION

The global surface observation network is dense

over lands and sparse over oceans. The missing data

over oceans is often filled with satellite observations,

but still to some limits. The satellite observations are

in good quality during the clear-sky day, but easily

distorted by severe weather. Ship observations are

another important data source. However, they are

restricted within the routine commercial routes to

cover certain parts of the ocean. Moreover, ship

observations fail to provide detailed surface

information for severe weather due to its detour for

safety. The real surface conditions underneath a

severe storm over ocean (e.g., typhoon or hurricane)

are still mysterious to us so far.

To overcome this problem, one may take

advantage of the state-of-the-art atmospheric and

oceanic numerical models developed in the recent

years. In other words, a numerical atmosphere-ocean

model system may be constructed for

simulating/predicting the atmospheric and oceanic

conditions associated with the severe weather event.

Such a numerical model simulation/prediction,

although not perfect, can at least provide us a

reasonable guess for surface conditions underneath

and near the storm over the open oceans. The

purpose of this study is to report the construction of

an atmosphere-ocean numerical model system. Its

performance in typhoon prediction is investigated.

280

2 THE PREDICTION SYSTEM

The model system constructed in this study aims at

predicting surface wind and sea wave in the western

North Pacific. This system consists of two models.

The atmospheric component is the Weather

Research and Forecasting (WRF) model (Skamarock

et al. 2005), while the oceanic component is the

NOAA WaveWatch version 3 (NWW3) model

(NOAA 2002). Given the NCEP aviation forecast as

the initial boundary condition, the WRF with the

145-km grid is first executed to make prediction for

the 80°-160°E, 10°S-50°N domain. Its predictions

are then dynamically downscaled by the second

WRF with a higher spatial resolution, 15-km grid, in

the 100°-150°E, 0°-35°N domain. Later, the 10-m

winds predicted by the second WRF are used as the

boundary conditions for the NWW3 model to

simulate sea wave, focusing on its height and

propagation. This NWW3 model is in a 0.2° grid and

has a spatial domain of 110°-145°E, 5°-35°N. The

model system is designed for predicting typhoon

activities over the western North Pacific.

3 THE MODEL PERFORMANCE

The prediction of Typhoon Bilis (No. 0604) is

selected as the case for examining the model

performance. Bilis formed over the west side of

Guam (141°E, 12°N) on July, 9, 2006, and moved

northwestward across Taiwan to decay over

southeast China on July, 15, 2006. Such a track is

reasonably predicted by the WRF (Fig. 1) in its

24- and 48- hr forecasts.

Fig. 1. The central position of Typhoon Bilis from observed,

initial condition, 24-hr prediction, and 48-hr prediction.

The averaged error in the spatial position of

typhoon center is 99 km in 24-hr forecast and 233

km in 48-hr forecast. These errors are in comparable

magnitudes with the official typhoon forecasts

conducted by weather services in Taiwan, Japan,

China, and USA, which range 94-110 km for 24-hr

forecast and 173-191 km for 48-hr forecast. The

prediction errors in the magnitude of 10-m winds

along the 125°E longitude (15°N, 20°N, 25°N, and

30°N) are evaluated against observations represented

by the NCEP Reanalysis data (Kalnay et al. 1996).

Table 1 reveals that prediction error in wind

magnitude is largest near the center of typhoon track

(20°N), followed by the dangerous semicircle

(typhoon’s right-hand side at 25°N and 30°N), and

smallest in the navigable semicircle (typhoon’s left-

hand side at 15°N).

Table 1. The averaged errors of 10-m wind magnitude for the

initial condition (0 hr), 24-hr prediction, and 48-hr prediction at

4 different meridional locations along the 125°E longitude

(unit: m/s).

bilis(125E) 0hr 24hr 48hr

15N 2.52 1.44 5.00

20N 5.60 4.75 9.35

25N 4.33 4.09 6.95

30N 4.85 3.02 4.57

The spatial pattern of 10-m winds delineate

clearly that wind speed is apparently larger in the

dangerous semicircle than in the navigable

semicircle (Fig. 2).

Fig. 2. The spatial pattern of predicted high wind (≧ 30 knots,

shading) at 10 m from the 24-hr prediction of the WRF model.

Sea level pressure is plotted in contours

Regarding the sea wave, the predicted height of

sea wave is validated against the wave analysis

issued by the Japan Meteorological Agency (JMA).

Its prediction errors (Table 2) are largest in the

dangerous semicircle, followed by the typhoon

center, and smallest in the navigable semicircle.

281

Table 2. As in Table 1, except for the averaged errors of the

height of sea wave (unit: m).

bilis(125E) 0hr 24hr 48hr

15N 0.98 1.02 1.61

20N 1.11 0.75 2.07

25N 1.83 2.69 3.46

30N 1.33 1.87 2.69

The prediction shows that the height of wave

(Fig. 3) in the dangerous semicircle can be twice as

large as that in the navigable semicircle (9 m vs.

4 m).

Fig. 3. The spatial pattern of predicted sea wave height

(shading) from the 24-hr prediction of the NWW3 model

4 SUMMARY

By investigating the temporal and spatial patterns of

the predicted 10-m winds and sea wave, general

performance of this model system in typhoon

prediction is summarized as follows:

− The numerical model system is able to reasonably

capture typhoon track in advance by 24 to 48 hrs.

− The dangerous regions with high wind (≧ 30

knots) and high sea (≧ 4 m) can be identified.

The spatial range of the prediction error (about

100-240 km) in typhoon center should be added

to broaden the spatial domain of dangerous

regions when the predictions are employed in

navigational plan.

− Contrast between the dangerous and navigable

semicircles of the typhoon are clearly portrayed

by this model system. The model prediction

reveals that spatial domain of high wind is much

larger in the dangerous semicircle than that in the

navigable semicircle, while the maximum height

of sea wave is about twice in magnitude in the

former than in the latter.

− The model prediction is able to largely simulate

the surface winds and sea wave over the oceans

underneath and near the typhoon. These surface

conditions are potential useful inputs for (1)

constructing the database for a navigational

simulator, (2) making evacuating plan for ship

from the harbor, and (c) issuing early warning for

navigation safety.

REFERENCES

Skamarock et al. 2005: A Description of the Advanced

Research WRF Version 2. NCAR Technical Note,

NCAR/TN-468+STR.

NOAA, 2002: “User manual and system documentation of

WAVE-WATCH-III version 2.22” edited by H. L. Tolman,

NOAA Technical Note.

Kalnay, E. & Coauthors, 1996: The NCEP/NCAR 40-year

Reanalysis Project. Bull. Amer. Meteor. Soc., 77, 437-471.