Conclusions 134

First, a review of model used to assess the hurricane wind hazard map for the ASCE-7 is carried out. The review indicated that currently the full track model and planetary boundary layer (PBL) wind field model are used as the basis to estimate the hurricane wind hazard which is mapped in the ASCE-7. The review also indicated that the typhoon wind hazard in China is often carried out using the circular subregion method, and that there are large discrepancies in the estimated return period value of the typhoon wind hazards in selected cities.

Four major tasks are carried out in this study: one focused on the hurricane wind hazard estimate in the US, and the remaining three focused on the typhoon wind hazard estimate in China. The first task provided the needed theoretical basis for the wind field modeling and track modeling used to estimate the typhoon wind hazard for mainland China.

The research carried out in this study contributed in the understanding of and knowledge for hurricane hazard estimation in several ways:

1) It is shown that the convection term in the governing equation for the wind field can be important and affect the wind field shape;

2) Adequate empirical full track model for west Northern Pacific basin can be developed; and

3) Well-documented typhoon wind hazard maps focused on wind engineering applications are developed coastal region of mainland China.

More specifically, from the results of the task focused on the estimation of the hurricane wind hazard in the US, it is concluded:

1) The PBL model given in Chow (1981) is solved in several studies (Cardone 1996, Vickery et al. 2000a) by neglecting the term u_cu_s in the moment equation. This

resulted in different wind field shapes and underestimates the maximum wind speed that is obtained by including such a term. The underestimation of the maximum wind speed is small (<3.5%) for typical value of p and B, but could be large for large value of p and B. The modeled wind field provide a closer resemblance to the wind fields in H*Wind if u_cu_s is considered.

2) A simplified track model is proposed. Its use leads to the estimated hurricane wind hazard similar to one used as the basis for the hurricane wind hazard shown in ASCE- 7.

3) Results from different combinations of wind field model and track model considered lead to relatively consistent hurricane wind hazard estimate, indicating the hurricane wind hazard model is robust.

From the results of tasks focused on the estimation of the typhoon wind hazard in the coastal region of the mainland China, it is concluded:

1) Typhoon wind hazard for nine major coastal cities in the mainland China is assessed by using the CSM combined with the PBL wind field model. Based on historical track dataset, probabilistic characterization is carried out for for key parameters including storm heading, translation velocity, and the central pressure difference by considering a circular region centered at each site of interest. It was indicated that the preferred probabilistic model for the central pressure difference could be site dependent.

2) The estimated wind hazards for nine cities are relatively insensitive to the size of the sub-region considered, but are affected by the adopted probability distribution model for p. The estimated wind hazard for Shanghai and Hong Kong is comparable to those based on surface wind observations. The code recommended return period values for several cities are lower than those obtained in this study by up to 8%. The large differences are for Ningbo and Wenzhou.

3) The spatially statistical characterizations of the key TC parameters are explored. The annual occurrence rate of TC for a considered circular region, is spatially varying, and decreases as the circle moves towards inland. The mean value of the heading

decreases from east to southwest; the standard deviation of the heading is relatively consistent as compared to the mean. The translation velocity tends to increase as the TC moves towards inland or from south to north. For modeling p, there are clearly identified zones where the lognormal (or the Weibull) distribution is preferred.

4) Contour maps for typhoon wind hazard are developed by using CSM combined with the PBL wind field model. The developed contour map for v50 is compared to that given in the design code. Regions outlined by contour lines ranging from 31 to 38 m/s in the developed contour map are comparable in size and shape to those in the contour map given by the design code; but regions outlined by contour lines ranging from 22 to 28 m/s in the developed contour map are located further inland compared to those in the contour map given in the design code. Sensitivity analysis shows that vT (T = 50- or 100-year return period) is not very sensitive to the radius of the circular. The consideration of spatially varying filling-rate model or the approximate model shown in Eq. (4.4) alters v100 by less than 6%; the use of lognormal or Weibull distribution to model p when using the CSM affect the estimated v100 by up to 12%.

5) An empirical full track model is developed to probabilistically predict the TC tracks from the genesis to lysis for western North Pacific basin. The spatially varying model coefficients are determined through regression analysis. Statistics of the key TC parameters estimated from the simulated tracks at kilometer posts compare favorably to those obtained from the historical tracks. The simulated tracks are used together with the PBL wind field model to estimate the typhoon wind hazard. The estimated vT compare well with those available in the literature or estimated based on the surface wind observations. The typhoon wind hazard maps for the coastal region assessed using the full track approach are presented and compared with those obtained based on the CSM. For sites that are overland and within 250 km from the coastline, the absolute relative difference between vT estimated by full track model and by the CSM with best fitted distribution for p is about 4.8% for T = 50- and 100-year.