A modular class of multisite monthly rainfall generators for water resource management and impact studies
A modular class of multisite monthly rainfall generators for water resource management and impact studies
复制标题
DOI：
10.1016/j.jhydrol.2012.07.043
复制
发表时间：
2012-09
影响因子：
6.4
通讯作者：
F. Serinaldi;C. Kilsby
中科院分区：
文献类型：
--
作者：
F. Serinaldi;C. Kilsby
本文介绍了一类随机多站点月降雨量发生器，用于水资源管理问题，如外部气候和非气候强迫机制下的干旱和极端降雨情景的敏感性分析。该模型框架依赖于三个要素：(1)基于对数变换观测的经典去季度化方案；(2)非参数Bootstrap重采样方法；(3)位置、尺度和形状的参数广义加性模型(GAMLSS)。由于Bootstrap和GAMLSS模块是每月模拟的替代技术，它们之间的自由选择使模型的结构模块化和灵活，因此它可以很容易地适应不同的气候条件，并可以根据具体的水资源问题进行定制。该模型被建立和校准，以模拟英格兰和威尔士六个地点的月降雨量，以便为干旱分析提供合适的输入。实例分析结果表明，该模型能较好地刻画降雨序列的几个特征。特别是，它能够模拟比观测到的更极端的低降水和高降水情景，以及再现年累积降雨量的分布，以及降水与北大西洋涛动(NAO)和海表温度(SST)等环流指数之间的关系，从而使该框架非常适合于在不同气候情景和附加强迫变量下进行敏感性分析。
This study introduces a class of stochastic multisite monthly rainfall generators devised for application in water resources management problems, such as the sensitivity analysis of droughts and extreme rainfall scenarios under external climatic and non-climatic forcing mechanisms. The modelling framework relies on three elements: (1) a classical deseasonalisation scheme based on log-transformed observations, (2) the nonparametric bootstrap resampling approach and (3) parametric Generalized Additive Models for Location, Scale and Shape (GAMLSS). As the bootstrap and GAMLSS modules are alternative techniques for simulating each month, the free choice between them makes the structure of the model modular and flexible, so that it can be easily adapted to different climatic conditions, and can be customized based on the specific water resource problem. The model was set up and calibrated to simulate monthly rainfall from six locations in England and Wales to produce a suitable input for drought analysis. The results of the case study point out that the model can capture several characteristics of the rainfall series. In particular, it enables the simulation of low and high rainfall scenarios more extreme than those observed as well as the reproduction of the distribution of the annual accumulated rainfall, and of the relationship between the rainfall and circulation indices such as North Atlantic Oscillation (NAO) and Sea Surface Temperature (SST), thus making the framework well-suited for sensitivity analysis under alternative climate scenarios and additional forcing variables.