基于空间贝叶斯层次模型的淮河流域气候极值特征分析

王怀军; 潘莹萍; 冯如

引用本文:	王怀军,潘莹萍,冯如.基于空间贝叶斯层次模型的淮河流域气候极值特征分析[J].灌溉排水学报,0,():-.
	WANG Huaijun,PAN Yingping,FENG Ru.基于空间贝叶斯层次模型的淮河流域气候极值特征分析[J].灌溉排水学报,0,():-.

【打印本页】【下载PDF全文】【查看/发表评论】【EndNote】【RefMan】【BibTex】

过刊浏览高级检索

本文已被：浏览 1689次下载 0次
分享到：微信更多字体:加大+\|默认\|缩小-
基于空间贝叶斯层次模型的淮河流域气候极值特征分析
王怀军¹, 潘莹萍², 冯如¹
1.淮阴师范学院城市与环境学院;2.北京师范大学地理科学学部

摘要:

【目的】验证空间贝叶斯层次模型在极端气候事件建模的适用性,分析淮河流域极端气候事件的空间分布规律【方法】基于空间贝叶斯层次模型,将经度、纬度与海拔作为模型协变量捕捉气候极值的空间变化特征。在建模过程中,将广义极值函数(GEV)作为其边际分布,采用马尔可夫链蒙特卡罗算法(MCMC)得到空间极值模型参数的后验分布。将该模型应用于淮河流域1960~2015年1日最大降水量(RX1day)和月极端最高气温(TXx)数据中,并将其导出的GEV参数和重现水平与最大似然法估计的参数进行对比。【结果】空间贝叶斯层次模型能够很好地估计观测站点GEV参数和重现水平。RX1day不同重现期重现水平从流域西北向东南增加；TXx重现水平具有典型的经向地带性,其重现水平从流域东部往西部增加。【结论】本文建立的空间极值模型可以获得没有观测台站所在位置的极端气候重现水平,该结果拓展了淮河流域极端气候事件时空规律研究。

关键词: 极端气候;贝叶斯层次模型;空间极值;淮河流域

DOI：

分类号:P467

基金项目:国家自然科学基金项目(41701034)、国家重点研发计划资助（2018YFC1508101）

Analysis of climate extremes by the spatial Bayesian hierarchical model in Huaihe River Basin, China

WANG Huaijun¹, PAN Yingping², FENG Ru¹

1.School of Urban and Environmental Sciences,Huaiyin Normal University,Huai’an;2.Faculty of Geographical Science,Beijing Normal University

Abstract:

【Objective】The objective of this study is to verify the applicability of the spatial Bayesian hierarchical model for simulating the spatial and temporal changes of extreme climate events【Method】 In this paper, a hierarchical Bayesian spatial model, named HKEVP, is applied to capture the spatial variation of the extremes with longitude, latitude and altitude as the covariates. The generalized extreme value distribution (GEV) was used as its marginal distribution. The Markov Chain Monte Carlo method (MCMC) was employed to infer the parameters of the hierarchical Bayesian model. The model is applied to the 55-year monthly maximum 1-day precipitation (RX1day) and monthly maximum value of daily maximum temperature (TXx) in the Huaihe River Basin, China. The GEV parameters and return levels derived from hierarchical Bayesian model were compared with that by the maximum likelihood estimates. 【Result】We found that the hierarchical Bayesian model behaved well in estimated GEV parameters of the unobserved locations. The RX1day return level at different return period increased from northwest to southeast of the basin; and the return level of TXx displayed a typical longitudinal zonality, with low values in the eastern basin and high values in the western basin.【Conclusion】The spatial extreme model established in this paper can obtain the return level of climate extremes of ungauged station. This result expands the study of the spatial and temporal climate extremes in the Huaihe River Basin.

Key words: Climate extremes; Bayesian hierarchical model; Spatial extremes; Huaihe River Basin