灾害情景是对灾害事件的未来多种结果及其结果实现的各种条件的描述。构建灾害情景需要进行灾害情景的模拟，包括致灾因子情景模拟、危险性情景模拟、风险情景模拟等。致灾因子情景模拟是灾害情景模拟的基础，可分为单致灾因子情景模拟和多致灾因子情景模拟。热带气旋是典型的多灾种自然灾害，成灾机制复杂，常见致灾因子包括大风、降水、风暴潮、海浪、洪水等。热带气旋大风易引发风暴潮、海浪等灾害，热带气旋降水易增加洪水风险，风暴潮、海浪在入海口地区与洪水叠加，大大提升灾害危险性。中国位于西北太平洋沿岸，是受到热带气旋影响最严重的国家之一，沿海地区热带气旋多致灾因子问题突出，组合情景复杂。热带气旋多致灾因子情景的模拟能为灾害模拟提供致灾因子情景依据，为灾害模型如洪水模型提供情景输入，也可用于次生灾害分析、应急预案编制、防灾减灾预演等方面。
本文面向热带气旋多致灾因子情景模拟需求，选取大风、降水、风暴潮及海浪作为热带气旋的关键致灾因子，首先，综述了目前热带气旋致灾因子情景模拟主要方法的研究现状。其次，收集了多种热带气旋致灾因子危险性数据。然后，基于现有可能最大热带气旋的情景设定方法，在可能最大热带气旋路径-强度中增加了降水参数的设定，研究了各等级可能最大热带气旋路径-强度-降水的情景设定方法，并在海口市海甸岛进行了案例应用。接着，以海口市东部海域为研究区，对热带气旋风暴潮-海浪年极值的二维联合概率进行了研究，计算了各区域二维情景发生概率。最后，以海口市南渡江入海口为例，分析了热带气旋降水-风暴潮-海浪的三维联合概率，构建了热带气旋降水-风暴潮-海浪三维相关关系，以可能最大热带气旋案例中总降水量为例，设定了降水-风暴潮-海浪三维组合情景，为多致灾因子情景模拟提供了参考依据。本文主要结论如下：
1）可能最大热带气旋路径-强度-降水可用于热带气旋多致灾因子情景模拟输入。分析各参数之间的关系，最大风速与中心最低气压、最大风速半径呈负相关关系；最大风速与小时总降水量的观测值存在明显负相关外包络线，与小时总降水量的拟合值呈正相关关系，两者看似关系相反，但由于较大的最大风速发生可能性更小，所以拟合值仍符合历史观测规律。在进行可能最大参数相关分析时，若相关性较差，可以考虑最大值的相关性；若关系较为复杂，可分析数据的二级不确定性。
2）面状区域不同位置的风暴潮-海浪的联合概率分析，可得到了不同位置不同风暴潮-海浪二维情景的发生概率大小，为进一步模拟风暴潮-海浪多致灾因子情景提供概率性依据。期中，风暴潮增减水高度和海浪有效波高的单因子拟合的最优边缘分布为GEV分布，可大范围用于研究区的Copula连接函数为Gumbel Copula。不同重现期下，风暴潮增减水高度和海浪有效波高有明显的地理分布趋势，风暴潮呈现越接近海岸线越高的趋势，海浪与其趋势相反，且单致灾因子分布趋势明显影响风暴潮-海浪二维联合分布概率的空间特征。
3）三维联合概率可以为多致灾因子模拟提供边界条件，以便更科学地评估多致灾因子综合作用。在热带气旋降水-风暴潮-海浪三维联合概率分析中，对小时总降水量、风暴潮增水高度、海浪有效波高的单致灾因子分布拟合效果较好的是GEV分布。降水-风暴潮、风暴潮-海浪、海浪-降水二维联合分布最优拟合Copula函数均为Frank Copula函数。Clayton Copula函数对于降水-风暴潮-海浪三致灾因子的联合概率分布拟合效果最佳。

A disaster scenario describes the possible results and realized conditions of a disaster event. The outputs of disaster scenario simulations usually include hazard scenarios, exposure scenarios, environmental scenarios, risk scenarios and so on. Simulations of hazard scenarios are the basis for the other scenario simulations, and can be divided into single-factor scenario simulation and multi-factor scenario simulation. Tropical cyclones (TCs) are typical events associated with multiple hazards, and the mechanism by which they cause cascading disaster losses could be very complex, which often bring multiple types of hazards to offshore and onshore areas, including wind, rainfall, storm surges, waves and riverine floods. The wind is easy to cause storm surge, wave and other disasters, the precipitation is easy to increase flood risk, and storm surge and wave may overlap with flood in the estuary area. These hazards usually interact with each other, and these interactions may cause greater amplified hazard intensities. China is located in the Northwest Pacific coast, which is one of the countries most seriously affected by TCs. The results of scenario simulation are expected to provide a basis for subsequent simulations multi-hazard scenarios and inputs for disaster model, and provide a basis for prepared responses and strategies for disasters.
In this paper, wind, precipitation, storm surge and wave were selected as the key hazards of TCs. Firstly, the latest progress on scenario simulation methods for TCs is reviewed. Secondly, the long-term time series of hazards are collected. Then, on the basis of the parameterization of Probable Maximum Tropical Cyclones (PMTC), added the setting of the total hourly precipitation of TCs, established the relationship between precipitation and maximum wind speed of PMTC at different scales, and carried out a case study in Haidian Island, Haikou City. Besides, taking the eastern area of Haikou City as the study area, the brivarite joint probability of annual maximum storm surge and wave is studied, and the occurrence probability of brivarite regional scenarios is calculated. Finally, taking the entrance of Nandu River in Hainan Province as an example, the trivariate joint probability of precipitation, storm surge and wave is calculated, and random scenarios are generated. Taking the total precipitation in the case of the PMTC as an example, a trivariate combination scenario of precipitation-storm surge-wave is set, which provides a reference for scenario simulation of multi hazards. 
The main conclusions of this paper are as follows:
1) On the basis of the parameterization of PMTC, the setting of the total hourly precipitation of tropical cyclone is added and then established the relationship between precipitation and maximum wind speed of PMTC at different scales, and then discussed the correlation and confidence interval of them. The maximum wind speed is negatively correlated with the minimum center pressure and the maximum wind speed radius. There is a negative correlation between the maximum wind speed and the precipitation observations, and a positive correlation between the maximum wind speed and the precipitation fitting values. The trend of their relationship seems to be opposite, but the larger the maximum wind speed is less likely to occur, so the fitting value still conforms to the law of historical observations. 
2) For scenario simulation of annual maximum storm surge and wave, GEV is the suitable distribution for the storm surge heights and significant wave heights, and the Gumbel copula is the best for fitting the joint probability distribution. When approaching the coastline, the storm surge heights become higher, while the wave heights become lower under different return periods. The trend of the marginal distribution affects the spatial features of the joint probability distribution, and the bivariate scenario has obvious spatial distribution regularities, which can reflect the bivariate joint probabilities at different locations under different scenarios.
3) For scenario simulation of precipitation, storm surge and wave, GEV is the best mariginal distribution for total hourly precipitation, strom surge heights and significant wave height. For precipitation-storm surge, storm surge-wave and wave-precipitation respectively, Frank Copula is the most suitable bivariate Copula. Besides, Clayton Copula is the optimal trivariate Copula for precipitation-storm surge-wave. Trivariate joint probability can provide boundary conditions for the simulation of multi-hazards, so as to evaluate the comprehensive effect of multi-hazards more scientifically.