DOI: 10.1007/s00382-015-2491-7
Scopus记录号: 2-s2.0-84946486880
论文题名: A new statistical precipitation downscaling method with Bayesian model averaging: a case study in China
作者: Zhang X. ; Yan X.
刊名: Climate Dynamics
ISSN: 9307575
出版年: 2015
卷: 45, 期: 2017-09-10 起始页码: 2541
结束页码: 2555
语种: 英语
英文关键词: Bayesian model averaging
; Monthly precipitation
; Multiple linear regression method
; Statistical downscaling
英文摘要: A new statistical downscaling method was developed and applied to downscale monthly total precipitation from 583 stations in China. Generally, there are two steps involved in statistical downscaling: first, the predictors are selected (large-scale variables) and transformed; and second, a model between the predictors and the predictand (in this case, precipitation) is established. In the first step, a selection process of the predictor domain, called the optimum correlation method (OCM), was developed to transform the predictors. The transformed series obtained by the OCM showed much better correlation with the predictand than those obtained by the traditional transform method for the same predictor. Moreover, the method combining OCM and linear regression obtained better downscaling results than the traditional linear regression method, suggesting that the OCM could be used to improve the results of statistical downscaling. In the second step, Bayesian model averaging (BMA) was adopted as an alternative to linear regression. The method combining the OCM and BMA showed much better performance than the method combining the OCM and linear regression. Thus, BMA could be used as an alternative to linear regression in the second step of statistical downscaling. In conclusion, the downscaling method combining OCM and BMA produces more accurate results than the multiple linear regression method when used to statistically downscale large-scale variables. © 2015, Springer-Verlag Berlin Heidelberg. 资助项目: NSFC, National Natural Science Foundation of China
; NSFC, National Natural Science Foundation of China
; NSFC, National Natural Science Foundation of China
Citation statistics:
资源类型: 期刊论文
标识符: http://119.78.100.158/handle/2HF3EXSE/53958
Appears in Collections: 过去全球变化的重建
There are no files associated with this item.
作者单位: College of Forestry, Shenyang Agriculture University, Shenyang, China; State Key Laboratory of Earth Surface Processes and Resource Ecology, Beijing Normal University, Beijing, China
Recommended Citation:
Zhang X.,Yan X.. A new statistical precipitation downscaling method with Bayesian model averaging: a case study in China[J]. Climate Dynamics,2015-01-01,45(2017-09-10)