| 杜奕良,涂新军,杜晓霞,陈晓宏,林凯荣,吴海鸥.基于高维Gaussian Copula函数的区域农田灌溉需水分析[J].水利学报,2018,49(3):323-331 |
| 基于高维Gaussian Copula函数的区域农田灌溉需水分析 |
| Analysis of regional irrigation water demand based on high-dimensional Gaussian Copula function |
| 投稿时间：2017-08-31 |
| DOI：10.13243/j.cnki.slxb.20170848 |
| 中文关键词: 灌溉需水 降水频率 高维联合分布 Gaussian Copula函数 典型年法 最大权函数 |
| 英文关键词: irrigation water demand frequency of precipitation high-dimensional joint distribution Gaussian Copula function typical year most-likely weight function |
| 基金项目:国家重点研发计划项目（2017YFC0405900）；国家自然科学基金项目（51479217） |
| 作者 | 单位 | E-mail | | 杜奕良 | 中山大学 水资源与环境系, 广东 广州 510275
广东省华南地区水安全调控工程技术研究中心, 广东 广州 510275 | | | 涂新军 | 中山大学 水资源与环境系, 广东 广州 510275
广东省华南地区水安全调控工程技术研究中心, 广东 广州 510275 | eestxj@mail.sysu.edu.cn | | 杜晓霞 | 中山大学 水资源与环境系, 广东 广州 510275
广东省华南地区水安全调控工程技术研究中心, 广东 广州 510275 | | | 陈晓宏 | 中山大学 水资源与环境系, 广东 广州 510275
广东省华南地区水安全调控工程技术研究中心, 广东 广州 510275 | | | 林凯荣 | 中山大学 水资源与环境系, 广东 广州 510275
广东省华南地区水安全调控工程技术研究中心, 广东 广州 510275 | | | 吴海鸥 | 中山大学 水资源与环境系, 广东 广州 510275
广东省华南地区水安全调控工程技术研究中心, 广东 广州 510275 | |
|
| 摘要点击次数: 2418 |
| 全文下载次数: 1635 |
| 中文摘要: |
| 以广东省农田灌溉需水为例，构建了分区降水的8维联合分布模型，提出了给定全省降水频率的全省及分区农田灌溉需水分析思路，采用同频、典型年和最大权函数3种方法对区域需水、分区降水频率及需水进行了分析计算和对比。结果显示：Gaussian Copula函数能够很好地模拟广东省8个分区年降水联合分布，单变量最优分布主要为广义极值分布和广义正态分布。3种方法之间的全省农田灌溉需水相差甚小，但是各分区需水差别相对较大。典型年法推求的分区降水频率差异最明显，最大权函数法推求的大部分分区降水频率处于同频位置附近，其中分区Ⅷ的降水频率在特枯和特丰时明显偏离同频位置，而且其置信区间范围明显大于其它分区，表明分区Ⅷ不确定性更大。由于基于高维Copula函数联合分布模拟既考虑了分区降水的独立分布，又考虑了它们之间的相依性，而且能够给出分区需水的置信范围，因此认为最大权函数法推求区域农田灌溉需水更为合理。 |
| 英文摘要: |
| In order to investigate irrigation water demand in Guangdong province, South China, as a study case, an eight-dimensional joint distribution of sub-regional precipitations was built. For given frequencies of precipitation of the entire province, the calculation procedures of irrigation water demand of the entire province and its sub-regions were proposed. Frequency combinations of sub-regional precipitation and water demands using three methods, i.e. the equalized frequency, the typical year and the most-likely weight function, were compared. The results demonstrate that the Gaussian copula function efficiently fits the joint distribution of precipitation of eight sub-regions, in which the generalized extreme value distribution and generalized normal distribution are the optimal univariate distribution for individual sub-regions. The differences of water demands of the entire province among the three methods are quite small, but those of individual sub-regions are comparatively large, in particular using the typical year method. The design frequencies of precipitation of individual sub-regions using the most-likely weight function mostly occur near the lines of the equalized frequencies,but those of the Ⅷ sub-region in the extreme dry and extreme wet conditions remarkably are far from the lines in association with a large uncertainty due to a wider confidence interval. The most-likely weight function method on the basis of the high-dimensional Gaussian copula can simulate individual univariate distributions of sub-regional precipitations and capture their dependences, and also present a confidence interval of sub-regional water demands, which is recommended to acquire irrigation water demand for a large-scale region. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|