Yu
Fujiang, Zhang Zhanhai
National Marine Environment Forecast Center
(No. 8 Dahuisi, Haidian District, Beijing 100081, China Tel. 010-62173615)
Lin
Yihua
LASG, Institute of Atmosphere Physics, Chinese Academy of Science
Abstract: In this paper, a storm surge numerical forecast system supported by GIS is built. This system consists in two subsystem. One is the storm surge numerical forecast system, in which multiple nested grid is used for improving the computational accuracy in coastal areas. The other is the geography information system. A case study of storm surge generated by typhoon Polly is carried out by using this system in Tianjin coastal area.
Keywords: storm surge model, forecast, inundation GIS
Among the littoral countries of Northwest Pacific Ocean, the hazards from storm surges in China is of highest frequency and most severe. The disastrous regions from typhoon surges distribute almost the whole coastal area. In recent years, there is an ascending trend in property damages from typhoon surges. The damages from typhoon surges for serious disaster years(1994,1996,1997) was over ten billion even several ten billion RMB. It is very important and necessary to study the forecasting techniques and mechanism of storm surge, as well as the strategy for disaster prevention in China. In order to predict storm surge accurately, a storm surge numerical forecast system supported by GIS will be built and a case study for Bohai Sea will be given.
The system is composed of two subsystem. One is the storm surge numerical forecast system, which is the key part. This system is responsible for calculation the total water level (storm surge plus astronomical tide) of entire model domain. The other is the geography information system (GIS), which is the accessorial subsystem. This system is responsible for view, statistics and query of the inundation area caused by storm surges. The main functions show in the Fig 1.
2.1.1 Basic equations
In geographical co-ordinates, the basic depth-averaged equations governing tide-surge motion may be written in the form

Fig. 1 The main functions of the storm surge numerical forecast system
(1)
(2)
(3)
where: t denotes time
q, j east-longitude and latitude
z elevation of the sea surface
u,v components of the depth-mean current
components of
, the wind stress on the sea surface
components of
, the bottom stress
atmospheric pressure on the sea
surface
D the total water depth
r the density of sea water, assumed uniform
R the radius of the Earth
g the acceleration of gravity
f the Coriolis parameter (f=2wsinj)
The above equations are closed by relating
bottom stress
, to the depth mean current, v, using a quadratic law
where k is a friction
parameter, and relating wind stress,
, to the wind velocity, w, also using a quadratic law
where
is the density of air and
is a drag coefficient. The
relationship between
and w suggested by Wu(1982)
covering conditions ranging from breeze to hurricane has been used here:
where w is in m/s.
2.1.2 Typhoon model
In this model, the atmospheric pressure distribution in a typhoon area is assumed to be represented by Fujita’s and Takahashi’s formula
where
is the central pressure of typhoon.
R is the radius of maximum wind of typhoon.
The basic wind field is assumed to be represented by Ueno Takeo’s formula(1981).
where
are the components of the speed of
typhoon in x and y directions.
2.1.3 Scheme and method of computation
ADI finite difference scheme with Arakawa c mesh is used to integrate equations (1) to (3).
It is well-know that in order to calculate the storm surge accurately, the domain of computation must be large enough to cover storm scale, which makes water boundary is more accurate. But when the surges propagate to shallow water region, for example, continental shelf, estuary and bay, the impact of coastal configuration and water depth is very important on surge. Hence, we have designed multiple grid system to calculate the storm surge in Bohai Sea. As shown in Table 1, the entire region is divided into three subregions, within each subregion difference time step and grid size are used.
Table 1 Grid system for storm surge numerical model
|
Region |
First |
Second |
Third |
|
Time step |
360sec |
120sec |
20sec |
|
Grid size |
1/10°, about 11Km |
1/30°, about 3.7Km |
1/180°, about 0.6Km |
|
No.of grid |
55´100, 5500 |
150´145, 21750 |
135´135, 18225 |
|
Longitude |
117°-127°E |
117°30’-122°20’E |
117°30’-118°15' E |
|
Latitude |
35°30’-41°N |
37°-41°N |
38°30'-39°15’N |
In the computation of the first grid system (coarse grid), initial conditions take the form of “cold start” condition
z=u=v=0 at time t=0
Coastal boundary condition is
where
is the component of current along
the outward-direction normal to the boundary. Open-sea boundary condition is a
“radiation” condition
where H is the depth of water, z is elevation of sea surface.
In the computation of the 2nd grid system (fine grid), coastal boundary is same as coarse grid, but the open-sea boundary is as follow:
where
is the elevation of sea surface
along the fine grid open-sea boundary which interpolated from coarse grid.
is the sum of the four main tidal
constituents of
, expressed as follows:
In the computation of the 3rd grid system (finest grid), open-sea boundary is
where
is the elevation of sea surface
along the finest grid open-sea boundary which interpolated by fine grid. As to
the coastal boundary which moves with the rising and falling surge, we used the
technique presented by Flather and Heaps (1975) to determine the ‘wet’ or
‘dry’ boundary.
The method of two grid system exchanging information is as follows:
(1) First and second grid system: The first grid system only gives the boundary value of 2nd grid system.
(2) Second and third grid system: Not only the 2nd grid system supplies the boundary value of the 3rd grid system, but also the new results of 3rd grid system update the old results of 2nd grid system. This method not only eliminates the paradise wave between the two grid system boundaries which makes computation stable, but also makes large grid system calculation more precise. In this method both large and small grid system must be calculated in the same time.
The storm surge model domain includes Tanggu and Hangu district of Tianjin in Bohai Bay. Therefore, 6 topography maps of 1:50000 scale were digitized include above areas. The digitized elements were topography, dikes, rivers, roads, salt farms and residential areas. The Arcview GIS software was used to build the system.
Tianjin is located in the western coastal area of Bohai Bay and it is a important industrial city and a commercial import-export harbor with a large population in the north of China. The Bohai Bay is very shallow. Hence, strong winds acting shallow water are the primary mechanism for the storm surge, in conjunction with the astronomical tides, can produce exceptional high water levels. The strong surge caused by No. 9216 was the most serious one stricken Tianjin since 1949. The maximum tide level is 5.93m which exceeded 1.23m of warning water level.
This model is used to simulate the surge generated by No. 9216 typhoon. Figure 2 shows the comparison between the computed and the observed surge elevations at Tanggu, tidal gauge station. We can see that the computed elevation curve looks quite similar to the observed one at each station, except for the first peak at Yangjiaogou tidal gauge station.
Note: The computed surges are coupled surges which is the residual of the total water level and the astronomical tide.
Fig. 2 The comparison between the computed and the observed surge at Tanggu
Fig. 3 shows the distributions of maximum flood range . The main inundation region is located along the estuary of Haihe River etc. which is in agreement with observed.
A storm surge numerical forecast system supported by GIS is built. This system consists in two subsystem. One is the storm surge numerical forecast system, in which multiple nested grid is used for improving the computational accuracy in coastal areas. The other is the geography information system. A case study of storm surge generated by typhoon Polly is carried out by using this system in Tianjin coastal area. The calculated results agree well with the observed ones. This indicates that this system is satisfactory for the forecasts of storm surge in Bohai Sea .
References
[1] Flather, R.A. and N.S. Heaps (1975) Tidal computations for Morecambe Bay. Geophysical Journal of the Royal Astronomical Society, 42,489-517.
[2] Yu Fujiang, Ye Lin and Wang Xinian, A High Resolution Storm Surge Prediction Model for Bohai Sea with Application to Typhoon 9216, Proceedings of The International Conference on Marine Disasters: Forcast and Reduction,China Ocean Press, 1998.
[3] Murty, T.S., R.A. Flather and R.F. Henry (1986) The storm surge problem in the Bay of Bengal, Progress in Oceanography, 16, 195-233.
[4] Jelesnianski, C.P. (1965) A numerical calculation of storm tides induced by a tropical storm on a continental shelf, Mon. Wea. Rev., 93, 343-358.

Fig. 3 The distributions of maximum flood of storm surge caused by typhoon Polly