A Bayesian decision approach to rainfall thresholds based flood warning
A Bayesian decision approach to rainfall thresholds based flood warning
复制标题
DOI：
10.5194/hess-10-413-2006
复制
发表时间：
2006-01-01
影响因子：
6.3
通讯作者：
Libralon, A.
中科院分区：
文献类型：
--
作者：
Martina, M. L. V.;Todini, E.;Libralon, A.
业务真实的时间洪水预报系统通常需要一个水文模型，以运行在真实的时间，以及一系列的水文信息学工具，以转换洪水预报为相对简单和明确的信息，决策者参与洪水防御。本文的范围是阐述在给定的河流部分提供洪水警报的基础上，直接比较的定量降水预报与临界降雨阈值的可能性，而不需要一个在线的真实的时间预报系统。这种方法导致一个非常简化的警报系统，供非技术利益相关者使用，也可以用来补充传统的洪水预报系统，在系统故障的情况下。临界降雨阈值，结合土壤水分的初始条件，结果从统计分析使用长水文时间序列结合贝叶斯效用函数最小化。在本文中，所提出的方法，筛河，在意大利的阿诺河支流的应用结果，给出了验证其实用性。
Operational real time flood forecasting systems generally require a hydrological model to run in real time as well as a series of hydro-informatics tools to transform the flood forecast into relatively simple and clear messages to the decision makers involved in flood defense. The scope of this paper is to set forth the possibility of providing flood warnings at given river sections based on the direct comparison of the quantitative precipitation forecast with critical rainfall threshold values, without the need of an on-line real time forecasting system. This approach leads to an extremely simplified alert system to be used by non technical stakeholders and could also be used to supplement the traditional flood forecasting systems in case of system failures. The critical rainfall threshold values, incorporating the soil moisture initial conditions, result from statistical analyses using long hydrological time series combined with a Bayesian utility function minimization. In the paper, results of an application of the proposed methodology to the Sieve river, a tributary of the Arno river in Italy, are given to exemplify its practical applicability.