热带气旋灾害是影响中国最严重的自然灾害之一，平均每年造成400余人死亡失踪(1951～2014年)，以及约473亿元直接经济损失(1985～2014年，2014年价格)。根据1951～2014年热带气旋死亡失踪人口数据，死亡失踪人数随时间呈现显著的减少趋势，但即便是21世纪以来，仍然有大量人员死亡的事件发生，在年度死亡失踪人口的下降趋势中表现出显著的波动性。根据1985～2014年热带气旋直接经济损失数据，中国年度经济损失在波动中呈不显著的上升趋势。热带气旋灾害人员伤亡和经济损失受热带气旋强度、人口和经济暴露性及脆弱性等因子的综合影响，驱动因素复杂。分析历史热带气旋死亡人口和经济损失的变化特征，探索其驱动因素，对理解历史损失变化规律，指导未来的风险防范措施具有重要的意义。 本文采用的研究方法为：首先重建了1951～2014年中国热带气旋大风、降水、人口，和1985～2014年土地利用、固定资本存量、国内生产总值(GDP)的空间分布数据，以及死亡人口、经济损失和防灾减灾措施的时间序列。其次，计算了改进后的大风、降水和风雨综合致灾强度，并采用Mann Kendall(MK)检验、数据拟合等统计方法分析了致灾强度的时空变化趋势。第三步针对热带气旋的主要承灾体类型，计算了人口、土地利用类型、固定资本存量和GDP的暴露特征，并分析其时空变化趋势。第四步从全国尺度分析热带气旋死亡人口变化的驱动因素，构建并计算了人口减灾能力、平均损失率、年度相对人口死亡率等人口脆弱性表征指标，并分析其变化特征，在此基础上使用负二项回归模型和标准差等统计方法，分析了中国热带气旋死亡人口长期减少趋势和波动性的主要驱动因素。第五步从全国尺度分析直接经济损失变化的驱动因素，选择了与直接经济损失相关性较好的GDP数据作为承灾体对象，构建了年度相对损失率作为经济脆弱性表征指标，并分析其变化特征，在此基础上通过分析各影响因素与直接经济损失之间的相关系数和显著性指标，研究各因素对直接经济损失变化的影响。第六步从省级尺度(以海南省为例)分析直接经济损失变化的驱动因素。基于省级经济损失脆弱性曲线一指数复合模型，反演了能够反映区域经济结构差异的经济脆弱性指标，并采用情景模拟分析方法，分析了致灾强度、暴露性、经济脆弱性曲线特征和土地利用结构特征等4类因素对直接经济损失变化的影响程度。 本文的主要结论包括以下6个方面： 1)本文构建的驱动因素统计分析方法和基于脆弱性曲线-指数复合模型的情景模拟方法，能够适用于中国和省级尺度热带气旋损失变化的驱动因素分析。 驱动因素统计分析方法，及其对应的平均损失率、年度相对损失率等脆弱性反演方法，对数据和模型精度要求不高，具有较强的通用性。该方法适用于研究区域广、区域内损失形成特征差异大、难以建立准确的损失评估模型的情况，如中国的死亡人口和直接经济损失变化驱动因素研究。 基于脆弱性曲线-指数复合模型的情景模拟方法建立了定量的经济损失评估模型，能够定量分析致灾强度危险性、暴露性和2类脆弱属性因素对经济损失变化的贡献率。其中的脆弱性曲线-指数复合模型设计了脆弱性曲线特征值和土地利用脆弱性指数2个脆弱性指标，能够弥补脆弱性曲线模型无法刻画暴露经济体内部结构差异的局限，提高了模型的模拟精度，并为定量分析各因素的影响程度提供了清晰的模型结构。该方法要求研究区域内的损失形成机制较均一，且有充足的样本数据，因此更适用于省级经济损失变化的驱动因素分析。 2)中国热带气旋致灾强度有显著的波动性，但没有显著的变化趋势。同时中国各地的致灾强度变化特征存在一定的区域差异。 1951～2014年间，影响中国的热带气旋大风、降水和风雨综合致灾强度指标及其年度累积值，均没有显著的变化趋势。但是低强度热带气旋的显著减少，使年总频次呈现显著减少趋势，并使中国热带气旋风雨综合致灾强度呈现不显著的下降趋势。不同的研究时段中，危险性变化特征也不同。1985～2014年间，中国的风雨综合致灾强度呈微弱但依然不显著的上升趋势，而海南省的风雨综合致灾强度则呈现不显著的下降趋势。 3)中国热带气旋死亡人口减少的主要原因是因工程和非工程措施改进而造成的人口脆弱性减小，暴露人口的显著增加起到反向作用。 1951～2014年热带气旋死亡人口呈现显著的下降趋势，其主要驱动因素是因工程和非工程措施改进而造成的人口脆弱性减小。虽然暴露人口的显著增加会增加死亡人口风险，但是它们对死亡人数的增加作用被脆弱性的降低抵消了。致灾强度的不显著下降有可能在一定程度上减少死亡人数，但不是死亡人口显著减少的主要原因。死亡人口的波动性主要由造成大量人员死亡的巨灾事件引起，巨灾事件主要由超出工程设防能力的高致灾强度引发，但缺乏预警预报以及不适当的应急反应行动等也是引发巨大生命损失的重要原因。 4)中国热带气旋直接经济损失增加的主要原因是经济暴露性的显著增加，经济脆弱性的显著下降起到反向作用。 1985～2014年中国热带气旋直接经济损失呈不显著的上升趋势，损失增加的最主要的原因是中国经济的快速发展造成的经济暴露性的显著增长。热带气旋强度的不显著上升趋势也会造成较小幅度的经济损失增加。虽然中国经济体整体的脆弱性因工程防护能力加强、建筑标准提高等原因而显著降低，但是它对经济损失的降低作用被其他因素的增加抵消了。当致灾强度超越工程设施设防水平时，容易造成巨大的直接经济损失，从而造成年度直接经济损失的明显波动。 5)中国各地区的直接经济损失变化驱动机制存在一定的区域差异，海南省经济脆弱性的升高是海南省直接经济损失增加的第二大影响因素。 1985～2014年海南省热带气旋直接经济损失呈不显著的上升趋势。采用驱动因素统计分析方法无法判断其经济损失变化的最主要驱动因素。采用情景模拟方法进行分析，认为海南省直接经济损失增加的最主要原因是暴露经济量的增长，因土地利用结构变化导致的脆弱性增加是直接经济损失增加的第二大原因，而致灾强度的不显著下降趋势削弱了直接经济损失的增长幅度，另外经济体自身的脆弱性曲线特征变化不大，因而对直接经济损失变化的贡献率也很小。可见，海南省的致灾强度和经济脆弱性变化方向与中国整体的变化方向相反，即中国各地的经济损失变化特征存在一定的区域差异。 6)加强非工程措施防范能力和合理规划建设用地的发展是未来减轻热带气旋灾害风险的关键措施之一。 面对未来仍有可能增多的超强热带气旋，任何工程措施都不能保证其万无一失。根据我国“以人为本”的灾害管理理念，为了减少未来超设防能力热带气旋带来的人员伤亡，应将加强非工程措施防范能力作为未来的优先发展事项，例如，继续加强预警预报能力、有效传递风险信息、不断完善应急预案等。对热带气旋经济损失而言，暴露经济量的快速增长和土地利用脆弱性的增加是直接经济损失增长的主要驱动因素，因此，合理规划热带气旋高风险区内的建设用地分布，如谨慎开展填海造地、减少侵占河湖漫溢区等措施，既能控制热带气旋经济暴露量的增长，又能降低土地利用脆弱性，是减轻热带气旋经济损失风险的关键措施之一。 本文的创新性体现在以下4个方面：1)充分利用热带气旋大风和降水的空间分布属性，构建了能够反映热带气旋影响范围的风雨综合危险性强度指标，并用于中国热带气旋危险性变化分析。2)针对区域人口、经济等结构复杂的承灾体，提出了通用性较强的脆弱性变化分析方法，包括年度相对损失率反演法和平均损失率统计法等。3)针对区域经济损失，构建了脆弱性曲线.指数复合模型。该模型能够同时刻画经济体对致灾强度的损失变化特征，以及暴露的土地利用结构差异对经济体损失率的影响。并构建了基于该模型的区域经济脆弱性指数和驱动因素情景模拟分析方法。4)阐明了中国热带气旋死亡人口和直接经济损失变化的主要驱动因素。中国热带气旋死亡人口减少的主要原因是因工程和非工程措施改进而造成的人口脆弱性减小，而中国热带气旋直接经济损失增加的主要原因是经济暴露性的显著增加。中国各地区的直接经济损失变化驱动机制存在一定的区域差异，海南省土地利用结构变化导致的经济脆弱性升高是海南省直接经济损失增加的第二大影响因素。 关键词：热带气旋，损失变化，驱动因素，致灾强度，暴露性，脆弱性，中国

 Tropical cyclone (TC) disasters are one of the most serious natural disasters in China, causing an average of more than 400 deaths and missing per year (1951-2014) and about 47.3 billion RMB in direct economic losses (1985-2014,2014 prices). According to the statistics of dead and missing people due to TCs from 1951～2014, the number of fatalities has been significantly decreasing over time. However, deadly TC events have still caused great losses of life even in the 21st century, which are characterized as significant abrupt fluctuations superimposed along the downward trend of the long-term fatality time series. According to the data on the direct economic losses of TCs from 1985 to 2014, China′s annual economic losses witnessed an insignificant upward trend in fluctuations. The deaths and economic losses of TC disasters are influenced by variables such as the intensity of TC hazards, the population exposed to TCs, and the vulnerability of people to TC hazards. It is thus of great significance to analyze their temporal characteristics and understand the forces driving these changes, which can help in understanding the causes of historical loss and guiding future risk prevention measures. The framework of this study is as follows: First, the time series of the TC wind, precipitation, spatial distribution of population, spatial distribution of land use, fixed capital stock and Gross Domestic Product (GDP) spatial distribution, fatality and economic losses caused by TC and disaster risk reduction (DRR) measures of China are reconstructed. Second, the improved power dissipation index, total precipitation, integrated intensity, exposed population and economic magnitude are calculated, and their changing trend and charictistics are analyzed. Third, the TC exposures of population, landuse area, fixed capital stock and GDP are calculated, and their changing trend and charictistics are analyzed. Forth, the main driving force of fatality change in country level is analyzed. Statistical methods such as negative binomial regression model and standard deviation statistics are used to analyze the driving force of the long-term change trend and fluctuations, based on the constructed population vulnerability indices, such as DRR capacity index, mean death rate and annual relative death rate. Fifth, the main driving force of economic loss change in country level is analyzed. Statistical methods such as correlation coefficient analysis and significance test are used to analyze the driving force of the long-term change trend and fluctuations, based on the analysis of annual relative loss rate as the economic vulnerability indice. Sixth, the main driving force of economic loss change in province level is analyzed, taking Hainan province as an example. A scenario simulation analysis based on a vulnerability curve--exponential composite model is constructed to analyze the driving factors of economic loss change in Hainan. The main conclusions include 6 aspects as follows: 1)The 2 driving force analysis methods constructed in this study, including the statistical analysis method and the scenario simulation method based on vulnerability curve-exponential composite model, can be applied at both country level and provincial level. The statistical analysis method, including its statistical vulnerability inversion methods, such as mean loss rate and annual relative loss rate, has low requirements on data and model accuracy and therefore has strong application availability. This method can be applied in the situation that it has different loss characteristics in different parts of the study area, and it is difficult to establish an accurate loss assessment model, such as the study of the driving factors of the loss change in China. A vulnerability curve-exponential composite model is constructed to identify the internal structural differences of TC exposure economies, and has been used to design a regional economic vulnerability inversion indicator and the scenario simulation analysis method for driving force analysis. The vulnerability curve-exponential composite model improves the accuracy of loss simulation and provides a clear model structure for quantitative analysis of the impact of 4 types of factors, such as hazard intensity, exposure, vulnerability curve characteristics and land use structure of exposed ecnomics. This method requires that the loss mechanism in the study area is relatively uniform and there are sufficient sample data, so it is more suitable for the dricing force analysis at provincial level. 2)The intensity of TCs impacting China shows no significant change trend but has significant volatility. And there are some regional differences in the change characteristics of TC intensity in different parts of China. From 1951 to 2014, there was no significant change trend in the wind intensity, precipitation intensity and the integrated intensity of TCs. However, the significant reduction of low-intensity TCs makes the annual frequency show a significant reduction trend, and makes the integrated intensity of TCs show a non-significant downward trend. In different research periods, the characteristics of intensity change are also different. From 1985 to 2014, the integrated intensity in China showed a upward but still not significant trend, while that of Hainan province showed a downward and non-significant trend. 3)The main driving force of the decreases in fatalities from 1951 to 2014 is the decrease in population vulnerability. Significant increases in the population exposure play a reverse From 1951 to 2014, the death population of TCs showed a significant downward trend. The decrease in vulnerability based on the improvement of structural and non-structural measures is the main driving force of the decreases in fatalities. Although the total population and exposure have increased dramatically in the coastal areas of China, their contributions to the increase in the fatality risk were counteracted by the decrease in vulnerability. The non-significant decrease in disaster intensity may reduce the number of deaths to some extent, but it is not the main reason for the significant decrease in the number of deaths. Abrupt and catastrophic disasters were mostly caused by TCs with hazards of high intensity that surpassed the capacity of structural measures; the lack of forecasting or early warning, as well as improper emergency response actions, may also have triggered the great loss of lives. 4)The main reason for the increase of direct economic losses of TCs in China is the significant increase of economic exposure. The significant decrease of economic vulnerability plays a reverse role. From 1985 to 2014, the direct economic losses of TCs in China showed non-significant upward trend. The main reason was the significant increase of economic exposure caused by the rapid development of China′s economy. An insignificant upward trend in TC intensity will also result in a smaller increase in economic losses. Although the overall vulnerability of the Chinese economy has been significantly reduced due to increased structural capabilities and improved building standards, its reduction in economic losses has been offset by increases in other factors. Abrupt and catastrophic disasters were mostly caused by TCs with hazards of high intensity that surpassed the capacity of structural measures, and induced the fluctuation of annual economic losses. 5)There are some regional differences in the driving mechanism of the change of direct economic loss in different regions of China. The increase of economic vulnerability in Hainan province is the second major factor to the increase of direct economic loss in Hainan province. From 1985 to 2014, the direct economic losses of TCs in Hainan province showed no significant upward trend. Statistical analysis of driving factors can not determine the main driving factors of economic loss changes. The results of scenario simulation show that the main reason for the increase of direct economic loss in Hainan province is the significant increase of economic exposure. The increase of vulnerability caused by the change of land use structure is the second major reason for the increase of direct economic loss, while the non-significant decrease of TC intensity weakens the increase of economic loss. It shows that the change direction of TC intensity and economic vulnerability in Hainan province is opposite to that of China as a whole, which means there are some regional differences in the characteristics of economic loss change in different parts of China. 6) The enhancement of non-structural measures and rational planning of construction land are some of the key measures to mitigate the risk of TC disasters in the future. With the possibility of more high intensity TCs in the future, no structural measures can ensure its safety. To reduce the fatalities of future TCs, especially those that may exceed the capacity of structural measures, the enhancement of non-structural measures and the adaptation of resilience strategies should be priorities for future people-centered disaster management. For example, we should continue to strengthen early warning and forecasting capabilities, effectively transmit risks information, and constantly improve emergency plans, etc. In order to reduce direct economic losses or slow down the growth rate, controlling the economic exposure and reducing the vulnerability of land use are both important to mitigate the risk of economic losses. Reasonable planning of the distribution of construction land in the highrisk areas of TC disaster, such as careful carry out land reclamation and reduce encroachment of rivers and lakes, can contribute to both reduce the economic exposure and the vulnerability. Therefore, reasonable planning of the distribution of construction land is one of the key measures to mitigate the risk of economic losses of TC disaster. The innovation of this study is embodied in the following 4 aspects: 1) Making full use of the spatial distribution attributes of TC wind and precipitation, an integrated TC event intensity index has been build and applied to the loss change analysis, so that it can simultaneously reflect the intensity and impact range of TCs. 2) A general vulnerability analysis method is proposed, including annual relative loss rate inversion method and mean loss rate statistics method. 3) In view of the internal structural differences of economic exposures, vulnerability curve-exponential composite model is constructed, which can simultaneously depict the characteristics of economic losses on TC intensity and the changes of land use structure. 4) The main driving factors for the change of the death population and direct economic losses of tropical cyclones in China are clarified. The main reason for the decrease of TC fatalities in China is the reduction of population vulnerability caused by the improvement of structual and non- structual measures, while the main reason for the increase of direct economic losses of TCs in China is the significant increase of economic exposure. There are some regional differences in the driving mechanism of the change of direct economic loss in China. The increase of economic vulnerability caused by the change of land use is the second major factor affecting the increase of direct economic loss in Hainan Province. KEY WORDS: Tropical cyclone, Loss change, Driving force, Hazard, Exposure, Vulnerability, China