海南省昌江黎族自治县王下乡具有良好的旅游资源和生态价值，而山洪灾害已成为制约王下乡旅游业可持续发展的重要因素之一。为了减轻自然灾害对旅游业的影响，必须对开发作出限制，但如果过于重视旅游景区自然灾害的风险防范，势必会对当地旅游景区的进一步开发造成阻碍。同时满足灾害经济损失最小、乡村旅游经济收入最高、环境最为生态宜居的规划方案，是王下乡未来旅游产业持续发展、安全发展的重要前提。

本文利用天气发生器、HEC-HMS和FLO-2D进行山洪灾害淹没模拟，得到王下乡山洪灾害危险性评估结果，再基于该结果以及乡村旅游经济收入和生态宜居评价，构建多目标优化数学模型，使用改进的遗传算法NSGA-Ⅱ求解非劣解集，提出了三个备选的规划方案，为完善王下乡的土地规划方案提供决策支持。

本文的主要结论如下：

（1）未来王下乡百年一遇山洪灾害的河流水深普遍超过5 m，三派村、俄力村、浪论村等主要村庄有较大的被淹没风险。

（2）构建灾害经济损失最小、乡村旅游经济收入最高、环境最为生态宜居3个目标函数，采用改进的遗传算法NSGA-Ⅱ求解多目标优化模型得到非劣解集，并选取了3套代表性的方案：方案甲——灾害经济损失最小、乡村旅游经济收入最低、环境最为生态宜居；方案乙——灾害经济损失中等、乡村旅游经济收入中等、环境较为生态宜居；方案丙，灾害经济损失最大、乡村旅游经济收入最高、环境最不生态宜居。

（3）在遭遇百年一遇降水引发的山洪灾害时，与原土地规划方案相比，方案甲平均每平方米的乡村旅游经济收入增加额大于灾害经济损失增加额，方案乙平均每平方米的乡村旅游经济收入增加额约等于灾害经济损失增加额，方案丙平均每平方米的乡村旅游经济收入增加额小于灾害经济损失增加额。综合3个目标的变化情况，方案甲更优。

（4）在遭遇百年一遇降水引发的山洪灾害时，甲、乙、丙三种方案的乡村旅游经济收入较原规划方案都有所提高；在三种备选方案之中，方案甲的灾害经济损失最低，因为其水域占比最高，很少有占用河道的现象。

Wangxia, Changjiang, Hainan has excellent tourism resources and ecological value, but flash floods have become one of the important factors restricting the sustainable development of tourism Restricting the development is necessary when reducing the impact of natural disasters on tourism. But if too much attention is paid to the risk prevention of natural disasters in tourist attractions, it will inevitably hinder the further development of local tourist attractions. A plan which simultaneously meeting the highest tourism economic income, the most livable ecology, and the minimum disaster losses is an important prerequisite for the sustainable and safe development of the tourism industry in Wangxia in the future.
In this study, a weather generator, HEC-HMS and FLO-2D were used to simulate the flood disaster in Wangxia to obtain the assessment results of flash flood disaster in Wangxia. Based on the results, and the evaluation of tourism economic benefits before disaster and ecological livability, a multi-objective optimization mathematical model was constructed, and the improved genetic algorithm NSGA-Ⅱ was used to solve the problem. Finally, three alternative plans are selected, which provide decision-making support for improving the land planning of Wangxia.
The conclusions are as follows:
(1) The water depth during the once-in-a-century flash flood in Wangxia generally exceeds 5 meters, and villages such as Sanpai, Eli, and Langlun are at high risk of being submerged. 
(2) Construct 3 objective functions: minimizing economic losses during disasters, maximizing rural tourism economic income, and making the environment the most ecologically livable. Use an improved genetic algorithm NSGA-II to solve the multi-objective optimization model to obtain the pareto front, and select three alternative plans: Plan A has the lowest economic losses, the lowest economic income, and the highest ecological livability; Plan B balances these 3 objectives; Plan C has the highest economic losses, the highest economic income, and the lowest ecological livability.
(3) Compared with the original village plan, the average rural tourism economic income of Plan A per square meter increased more than economic losses during disasters, the average rural tourism economic income of Plan B per square meter increased approximately equal to economic losses during disasters, and the average rural tourism economic income of Plan C increased less than economic losses during disasters. Plan A is more optimal. 
(4) The 3 alternative plans have all increased their total rural economic income compared to the original village plan. Among the three alternative plans, the economic losses during disasters of Plan A is the lowest, as its water area accounts for the highest proportion and there are few occurrences of occupying the river channel.