基于信息扩散理论的四川盆地农业旱灾风险评估

Risk assessment of drought on agricultural land in the Sichuan Basin based on information diffusion theory

采用四川盆地农业旱灾灾情资料数据,基于信息扩散理论,研究了不同风险水平下的农业旱灾发生的概率及其分布,并对四川盆地的农业旱灾风险进行区划。结果表明:正态信息扩散方法可以较好地拟合农业旱灾受灾率的风险分布。与20世纪末相比,21世纪初四川盆地的眉山、重庆等地区发生受灾率xi≥10%和xi≥40%的农业旱灾风险增加。以受灾率xi≥5%和xi≥40%的超越概率作为农业旱灾风险区划阈值,四川盆地农业旱灾风险表现出区域性差异,农业干旱灾害高风险区位于巴中、中江、泸县等地区;西部地区发生农业干旱灾害的风险较小,与该地区丰富的降水量以及平原地形有关。该风险区划结果较为合理,可为当地政府防范农业旱灾提供决策依据。

Natural disasters result from the complex relationships and interactions of the vulnerability of disaster-bearing bodies with the disaster-causing factors.The nature of natural disaster risk is described by the probability distribu- tion which is associated with the magnitude and duration of the natural disaster.Evaluating a risk involves establis- hing a scientifically robust method to estimate the probability magnitude of a natural disaster risk.The mathemati- cal theory of natural disaster risk assessment is essentially equivalent to an issue of probability and statistics, which can be described by the method of probability.The usual sample size required in probability-based statistical analyses is at least 30.However,in disaster risk assessment, especially when small areas are adopted as the evaluation unit, the data collection period is often short and statistics are incomplete. Therefore, it is difficult to reflect the variation in the risk of a disaster over certain study areas and big bias of computed results from observation often happen if we use traditional statistical methods to assess risk. Drought is one of the most serious meteorological disaster and often causes considerable socioeconomic losses in the affected region.The Sichuan basin China is a very important agricultural production area.Researching and assessing the risk of agricultural drought in this region is helpful in making service decisions and carries huge social significance.However,risk assessment work on agricultural drought in the Sichuan basin has been minimal, and mainly focused at the city prefecture level. Agricultural disaster area is related to the disaster strength and scale, thus, the assessment of agricultural drought risk can be accomplished via a method ＂disaster loss evaluation＇. Based on agricultural drought disaster data in Sichuan basin during 1984--2007 and information diffusion theory, the probability of agricultural drought, and its spatial distribution, are discussed under different risk levels. Accordingly,the risk of agricultural drought disaster in the Sichuan basin is separated （spatially） into a few of zones.Results show that the normal information diffusion method can provide a good fit for the agricultural drought disaster rate which is a proportion of the disaster area to the cultivated one.Further investigation shows that the difference in risk between Chongqing and Meishan,where the agricultural drought disaster rate is greater than or equal to 10% and 40% ,respectively,increasing from the end of the 20th century to the beginning of the 21st century.An exceeding probability means an accumulated probability when the drought disaster rate is larger than a threshold,which is defined as 5% and 40% in this study,respectively.Cities such as Bazhong, Zhongjiang and Luxian are found to be high-risk regions.Low-risk regions are located in the western Sichuan basin,owing to its abundant precipitation. The research provides some useful decision-making information for those involved in disaster reduction and emergency management. This study shows that small samples of disaster data can be applied in natural disaster risk assessment by u- sing the information diffusion method. The accuracy and completeness of assessment data is probably the most crucial aspect in evaluating results.Due to the disaster data record being subjective and the sown area of the crop not completely matching the data in time, the result is not completely satisfactory.The method is a time-independ- ent disaster risk assessment method that can be used to carry out statistical analysis of the disaster risk for a region where disaster events have occurred, and can also provide an indication of the natural disaster risk level of the re- gion.Further research is necessary to help us make effective forecasts of long-term drought risk in the future.