地理科学进展 ›› 2016, Vol. 35 ›› Issue (3): 339-348.doi: 10.18306/dlkxjz.2016.03.008

• 研究论文 • 上一篇    下一篇

雅鲁藏布江流域不同源降水数据质量对比研究

黄浠1,2(), 王中根1,*(), 桑燕芳1, 杨默远1,2, 刘晓聪1,2, 巩同梁3   

  1. 1. 中国科学院地理科学与资源研究所,陆地水循环及地表过程重点实验室,北京 100101
    2. 中国科学院大学,北京 100049
    3. 西藏自治区水利厅,拉萨 850000
  • 收稿日期:2015-07-01 接受日期:2015-11-01 出版日期:2016-03-25 发布日期:2016-03-25
  • 通讯作者: 王中根 E-mail:huangx.13s@igsnrr.ac.cn;wangzg@igsnrr.ac.cn
  • 作者简介:

    作者简介:黄浠(1991-),女,湖南岳阳人,硕士研究生,主要从事水文水资源研究,E-mail: huangx.13s@igsnrr.ac.cn

  • 基金资助:
    国家自然科学基金重点项目(41330529);中国科学院战略性先导科技专项(XDB03030202);西藏自治区水资源承载力研究项目

Precision of data in three precipitation datasets of the Yarlung Zangbo River Basin

Xi HUANG1,2(), Zhonggen WANG1,*(), Yanfang SANG1, Moyuan YANG1,2, Xiaocong LIU1,2, Tongliang GONG3   

  1. 1. Key Laboratory of Water Cycle and Related Land Surface Processes, Institute of Geographic Sciences and Natural Resources Research, CAS, Beijing 100101, China
    2. University of Chinese Academy of Sciences, Beijing 100049, China
    3. Water Resources Department of Tibet Autonomous Region, Lhasa 850000, China
  • Received:2015-07-01 Accepted:2015-11-01 Online:2016-03-25 Published:2016-03-25
  • Contact: Zhonggen WANG E-mail:huangx.13s@igsnrr.ac.cn;wangzg@igsnrr.ac.cn
  • Supported by:
    Key Program of the National Natural Science Foundation of China, No.41330529;Strategic Priority Research Program of the Chinese Academy of Sciences, No.XDB03030202;Program on Water Carrying Capacity of Tibet Autonomous Region

摘要:

本文以雅鲁藏布江流域为研究区,利用13个气象站点的实测降水量数据在年和月尺度上验证了中国地面降水网格数据、CRU(Climatic Research Unit)降水数据和GLDAS(Global Land Data Assimilation System)降水数据的精度,并分析了不同源数据降水量年际变化特征和概率分布特性之间的差异。结果表明:4种不同来源的降水数据均存在一定程度的差异。年尺度和月尺度上中国地面降水网格数据与实测降水量数值最接近;而CRU降水数据和GLDAS降水数据与实测降水量相差较大,在使用时需谨慎。从空间差异性看,年尺度上CRU降水数据在每个站点与实测降水数据的相关性均高于GLDAS降水数据,说明前者的空间一致性较好,但相对误差却比GLDAS降水数据大。从年内变化趋势看,中国地面降水网格数据能较好地反映流域降水月尺度的变化特征,CRU降水数据则在流域大部分地区的汛期时段都存在明显的高估,而GLDAS数据无法反映月降水变化趋势,年内坦化现象十分显著。从年际变化特征看,中国地面降水网格数据能较好地反映实际降水量的年际变化特征,而GLDAS降水数据和CRU降水数据反映的降水量年际变化特征偏小,其中GLDAS数据的坦化现象更严重,会高估低降水值,低估高降水值。从降水概率分布情况来看,3种来源的降水数据均不能反映站点实测的极端降水事件。

关键词: 雅鲁藏布江, 降水, 数据挖掘, 时空变化, 概率分布

Abstract:

The Yarlung Zangbo River is a transboundary river with rich hydropower resources. Reliable precipitation data are important for water resource development planning of the region. Due to the high elevation, complex topography, and severe climate, especially in the western part of the basin, however, rainfall stations are sparse. Precipitation estimation from satellite data or assimilation provides potential alternatives for precipitation measurements in regions where conventional precipitation gauges are not readily available. In this study, the performance of the gridded Chinese ground precipitation dataset, the Climatic Research Unit (CRU) precipitation dataset, and the precipitation data of the Global Land Data Assimilation System (GLDAS) in 1973-2013 were evaluated for the Yarlung Zangbo River Basin using observations from 13 meteorological stations. The results show that the four precipitation datasets significantly differ. The annual gridded Chinese ground precipitation dataset is the closest to the observed data while CRU and GLDAS precipitation datasets should be calibrated before use due to their limited precision. The CRU precipitation data show strong correlation with the observed precipitation, which indicates that there is a relatively high consistency between the CRU precipitation dataset and observed precipitation although its mean relative error is large. Monthly data analysis shows that the gridded Chinese ground precipitation dataset can reflect the variation characteristics while the CRU precipitation dataset tends to overestimate in flood season. Different from these two datasets, the GLDAS precipitation dataset presents obvious smoothing effect during the year. Annual variation of precipitation in the gridded Chinese ground precipitation dataset is closer to that of the observed precipitation while the coefficients of variation of precipitation in the other two datasets are much smaller. The GLDAS dataset overestimates precipitation in drier areas and underestimate precipitation in areas where annual precipitation is high. All the three precipitation datasets are unable to reflect the extreme precipitation events according to the probability distribution. The probability distribution of the GLDAS dataset concentrates in the range of 300~500 mm while the probability distribution of CRU precipitation ranges from 200~500 mm, higher than the observed precipitation.

Key words: Yarlung Zangbo River, precipitation, data mining, spatiotemporal variation, probability distribution