高级检索

留言板

尊敬的读者、作者、审稿人, 关于本刊的投稿、审稿、编辑和出版的任何问题, 您可以本页添加留言。我们将尽快给您答复。谢谢您的支持!

姓名
邮箱
手机号码
标题
留言内容
验证码

PSO-PSR-ELM集成学习算法在地面气温观测资料质量控制中的应用

张颖超 姚润进 熊雄 沈云培

张颖超, 姚润进, 熊雄, 沈云培. PSO-PSR-ELM集成学习算法在地面气温观测资料质量控制中的应用[J]. 气候与环境研究, 2017, 22(1): 59-70. doi: 10.3878/j.issn.1006-9585.2016.16013
引用本文: 张颖超, 姚润进, 熊雄, 沈云培. PSO-PSR-ELM集成学习算法在地面气温观测资料质量控制中的应用[J]. 气候与环境研究, 2017, 22(1): 59-70. doi: 10.3878/j.issn.1006-9585.2016.16013
Yingchao ZHANG, Runjin YAO, Xiong XIONG, Yunpei SHEN. Application of PSO-PSR-ELM-Based Ensemble Learning Algorithm in Quality Control of Surface Temperature Observations[J]. Climatic and Environmental Research, 2017, 22(1): 59-70. doi: 10.3878/j.issn.1006-9585.2016.16013
Citation: Yingchao ZHANG, Runjin YAO, Xiong XIONG, Yunpei SHEN. Application of PSO-PSR-ELM-Based Ensemble Learning Algorithm in Quality Control of Surface Temperature Observations[J]. Climatic and Environmental Research, 2017, 22(1): 59-70. doi: 10.3878/j.issn.1006-9585.2016.16013

PSO-PSR-ELM集成学习算法在地面气温观测资料质量控制中的应用

doi: 10.3878/j.issn.1006-9585.2016.16013
基金项目: 

国家自然科学基金项目 Grant 41675156

江苏省六大人才高峰项目 Grant WLW-021

详细信息
    作者简介:

    张颖超,男,1960年出生,教授,主要从事气象资料质量控制、气象灾害评估与损失预测、气象服务效益等研究。E-mail:yc.nim@163.com

    通讯作者:

    姚润进,E-mail:runjin_1991@163.com

  • 中图分类号: P413

Application of PSO-PSR-ELM-Based Ensemble Learning Algorithm in Quality Control of Surface Temperature Observations

Funds: 

National Natural Science Foundation of China Grant 41675156

Six Talent Peaks Project in Jiangsu Province Grant WLW-021

  • 摘要: 针对台站密度低、新建台站以及特种单要素站等无法获得有效邻站或内部参考资料情况下的质量控制问题,从气温时间序列的混沌特性出发,考虑气温在短时间内的连续性和稳定性,提出一种基于粒子群(ParticleSwarm Optimization,PSO)改进的相空间重构法(Phase Space Reconstruction,PSR)和极限学习机(Extreme LearningMachine,ELM)的集成学习算法的地面逐时气温观测资料的单站质量控制方法,实现气温资料的质量控制。为检验该方法的适用性,运用该方法对江苏省八市2007~2009年的地面气温观测资料进行质量控制,并与传统单站方法及切比雪夫多项式内插法(Tshebyshev Polynomial Interpolation,TPI)进行对比。实验结果表明,该方法相比较于TPI和传统方法可以更有效地标记出可疑数据,具有检错率高、地区和气候适应性、可控性强等优点。
  • 图  1  PSO 算法改进PSR和ELM 组合算法流程图

    Figure  1.  The flow chart of combination algorithm based on PSR(Phase Space Reconstruction)and ELM(Extreme Learning Machine)improved by PSO(Particle Swarm Optimization)

    图  2  PSO 改进前、后PSR-ELM 算法对2009 年南京站气温序列三种预测指标(a)RMSE、(b)MAE、(c)NSC 的对比

    Figure  2.  Comparisons of the three predictive indices(a)RMSE, (b)Mean Absolute Error(MAE), and(c)Nash-Sutcliffe model efficiency Coefficient(NSC)from PSR-ELM algorithm before and after improved by PSO for temperature at Nanjing Station in 2009

    图  3  2007~2009 年南京站两种资料(预测资料和台站ERA-Interim 再分析资料)与基准资料对比的(a)均方根误差和(b)相关系数曲线

    Figure  3.  RMSE and(b)correlation coefficients between forecast data,station ERA-Interim reanalysis data and standard data from 2007 to 2009 at Nanjing station

    图  4  2007 年、(b)2008 年、(c)2009 年南京等八市台站两种资料的年均偏差分布

    Figure  4.  Distributions of annual average deviations of forecast data and station ERA-Interim reanalysis data from(a)2007,(b)2008,and(c)2009 at observationstations of Nangjing(NJ),Xuzhou(XZ),Lianyungang(LYG),Huai’an(HA),Yancheng(YC),Yangzhou(YZ),Nantong(NT),and Wuxi(WX)

    图  5  2007 年、(b)2008 年、(c)2009 年PSO-PSR-ELM 方法、TPI 方法、传统方法在南京站检错效果对比

    Figure  5.  Comparisons of detecting ratios of PSO-PSR-ELM, TPI(Tshebyshev Polynomial Interpolation), and conventional single-station quality control method(CSSQCM)for observations from(a)2007, (b)2008,and(c)2009 at Nanjing station

    图  6  南京等八市(a)2007 年、(b)2008 年、(c)2009 年气温的PSO-PSR-ELM 方法与TPI 方法及传统方法检错效果对比

    Figure  6.  Comparisons of detecting ratios of PSO-PSR-ELM, TPI, and conventional single-station quality control method for temperature from(a)2007, (b)2008, and(c)2009 at observation stations of Nangjing(NJ), Xuzhou(XZ), Lianyungang(LYG), Huai'an(HA), Yancheng(YC), Yangzhou(YZ), Nantong(NT), and Wuxi(WX)

    图  7  南京站2007~2009 年PSO-PSR-ELM 预测(a)均方根误差、(b)检错率、(c)质控耗时与可控步长的关系

    Figure  7.  Relationships between(a)RMSE, (b)detecting ratio,(c)consuming time of the PSO-PSR-ELM method and controllable time step for observationsfrom 2007 to 2009 at Nanjing station

    表  1  2007~2009 年八市观测站气温时间序列的最大Lyapunov 指数

    Table  1.   The maximum Lyapunov exponents of the temperature time series collected from observation stations in eight cities from 2007 to 2009

    城市 嵌入维数 延迟时间/h 最大Lyapunov指数
    南京 15 2 0.0708
    徐州 16 3 0.0191
    连云港 25 2 0.0112
    淮安 26 3 0.0376
    盐城 30 4 0.0051
    扬州 19 4 0.0076
    南通 21 5 0.0043
    无锡 24 3 0.0233
    下载: 导出CSV

    表  2  八市 2007~2009 年PSO-PSR-ELM 质量控制法的检验效果

    Table  2.   Testing results with the quality control method ofPSO-PSR-ELM in eight cities from 2007 to 2009

    城市 检错率
    2007 2008 2009
    南京 81.02% 83.12% 84.02%
    徐州 82.17% 80.68% 82.45%
    连云港 80.02% 79.30% 81.28%
    淮安 83.15% 82.52% 85.19%
    盐城 80.16% 83.13% 82.06%
    扬州 84.53% 81.24% 85.19%
    南通 81.09% 83.73% 84.02%
    无锡 83.71% 82.16% 83.23%
    下载: 导出CSV

    表  3  2007~2009 年四市在4 个季度下最佳f 及季度平均检错率

    Table  3.   The optimum f(quality control parameter)values and corresponding detecting ratios in the four seasons from 2007 to 2009 at observation stations in four cities Nangjing, Xuzhou, Wuxi, and Huai'an

    年份 季度 南京 徐州 无锡 淮安
    f 检错率 f 检错率 f 检错率 f 检错率
    2007 1 1.5 80.41% 1.5 85.64% 1.5 84.79% 1.5 84.56%
    2 1.0 81.37% 1.5 80.18% 1.0 80.06% 1.5 83.64%
    3 1.5 78.93% 1.5 80.71% 2.0 78.98% 2.0 81.79%
    4 1.5 82.49% 1.5 81.36% 1.5 82.17% 1.5 80.93%
    2008 1 2.2 81.45% 1.5 85.89% 1.5 82.86% 1.5 85.57%
    2 1.0 79.69% 1.5 78.35% 1.5 79.98% 1.5 81.85%
    3 2.0 80.02% 2.0 78.03% 2.0 78.79% 2.0 84.86%
    4 1.5 83.05% 1.5 83.15% 1.5 81.30% 1.5 80.05%
    2009 1 1.5 79.90% 1.5 88.23% 1.5 85.15% 1.5 83.38%
    2 1.0 81.14% 1.5 79.92% 1.5 77.42% 1.5 84.47%
    3 2.0 75.01% 2.0 77.31% 2.0 80.41% 2.0 84.82%
    4 1.5 86.36% 1.5 80.30% 1.5 80.30% 1.5 83.33%
    下载: 导出CSV
  • [1] 陈帝伊, 柳烨, 马孝义. 2012. 基于径向基函数神经网络的混沌时间序列相空间重构双参数联合估计[J]. 物理学报, 61(10):100501. http://www.cnki.com.cn/Article/CJFDTOTAL-WLXB201210005.htm

    Chen Diyi, Liu Ye, Ma Xiaoyi. 2012. Parameter joint estimation of phase space reconstruction in chaotic time series based on radial basis function neural networks[J]. Acta Physica Sinica (in Chinese), 61(10):100501. http://www.cnki.com.cn/Article/CJFDTOTAL-WLXB201210005.htm
    [2] 丁一汇. 2009. 全球气候变化中的物理问题[J]. 物理, 38(2):71-83. doi: 10.3321/j.issn:0379-4148.2009.02.001

    Ding Yihui. 2009. Physical problems in the global climate change[J]. Physics (in Chinese), 38(2):71-83, doi:10.3321/j.issn:0379-4148.2009. 02.001.
    [3] 段晚锁, 丁瑞强, 周菲凡. 2013. 数值天气预报和气候预测可预报性研究的若干动力学方法[J]. 气候与环境研究, 18(4):524-538. doi: 10.3878/j.issn.1006-9585.2012.12009

    Duan Wansuo, Ding Ruiqiang, Zhou Feifan. 2013. Several dynamical methods used in predictability studies for numerical weather forecasts and climate prediction[J]. Climatic and Environmental Research (in Chinese), 18(4):524-538, doi: 10.3878/j.issn.1006-9585.2012.12009.
    [4] 窦以文, 屈玉贵, 陶士伟, 等. 2008. 北京自动气象站实时数据质量控制应用[J]. 气象, 34(8):77-81. http://www.cnki.com.cn/Article/CJFDTOTAL-QXXX200808013.htm

    Dou Yiwen, Qu Yugui, Tao Shiwei, et al. 2008. The application of quality control procedures for real-time data from automatic weather stations[J]. Meteorological Monthly (in Chinese), 34(8):77-81. http://www.cnki.com.cn/Article/CJFDTOTAL-QXXX200808013.htm
    [5] Eberhart R C, Kennedy J. 1995. A new optimizer using particle swarm theory[C]//Proceedings of the Sixth International Symposium on Micro Machine and Human Science. Nagoya, Japan:IEEE, 1:39-43, doi: 10.1109/MHS.1995.494215.
    [6] Feng S, Hu Q, Qian W H. 2004. Quality control of daily meteorological data in China, 1951-2000:A new dataset[J]. International Journal of Climatology, 24(7):853-870, doi: 10.1002/joc.1047.
    [7] 高路, 郝璐. 2014.ERA-Interim气温数据在中国区域的适用性评估[J]. 亚热带资源与环境学, 9(2):75-81. doi: 10.3969/j.issn.1673-7105.2014.02.012

    Gao Lu, Hao Lu. 2014. Verification of ERA-Interim reanalysis data over China[J]. Journal of Subtropical Resources and Environment (in Chinese), 9(2):75-81, doi:10.3969/j.issn. 1673-7105.2014.02.012.
    [8] 高志刚, 骆敬新, 刘克修, 等. 2015.ERA-Interim再分析数据在中国沿海的质量评估[J]. 海洋科学, 39(5):92-105. doi: 10.11759/hykx20140522001

    Gao Zhigang, Luo Jingxin, Liu Kexiu, et al. 2015. Evaluation of ERA-Interim reanalysis data along coast of China[J]. Marine Sciences (in Chinese), 39(5):92-105, doi: 10.11759/hykx20140522001.
    [9] Grassberger P, Procaccia I. 1983. Measuring the strangeness of strange attractors[J]. Physica D, 9(1-2):189-208, doi: 10.1016/0167-2789(83)90298-1.
    [10] 韩海涛, 李仲龙. 2012. 地面实时气象数据质量控制方法研究进展[J]. 干旱气象, 30(2):261-265. doi: 10.3969/j.issn.1006-7639.2012.02.018

    Han Haitao, Li Zhonglong. 2012. Review of quality control methods of surface real time meteorological data[J]. Journal of Arid Meteorology (in Chinese), 30(2):261-265, doi: 10.3969/j.issn.1006-7639.2012.02.018.
    [11] Huang G B, Zhu Q Y, Siew C K. 2006. Extreme learning machine:Theory and applications[J]. Neurocomputing, 70(1-3):489-501, doi: 10.1016/j.neucom.2005.12.126.
    [12] Hubbard K G, You J S. 2005. Sensitivity analysis of quality assurance using the spatial regression approach-A case study of the maximum/minimum air temperature[J]. J. Atmos. Oceanic Technol., 22(10):1520-1530, doi: 10.1175/JTECH1790.1.
    [13] Hubbard K G, Guttman N B, You J S, et al. 2007. An improved QC process for temperature in the daily cooperative weather observations[J]. J.Atmos. Oceanic Technol., 24(2):206-213, doi: 10.1175/JTECH1963.1.
    [14] Ingleby N B, Lorenc A C. 1993. Bayesian quality control using multivariate normal distributions[J]. Quart. J. Roy. Meteor. Soc., 119(513):1195-1225, doi: 10.1002/qj.49711951316.
    [15] Kennel M B, Abarbanel H D I. 2002. False neighbors and false strands:A reliable minimum embedding dimension algorithm[J]. Physical Review E, 66(2):026209, doi: 10.1103/PhysRevE.66.026209.
    [16] Kubecka P. 2001. A possible world record maximum natural ground surface temperature[J]. Weather, 56(7):218-221, doi:10.1002/j.1477-8696. 2001.tb06577.x.
    [17] Lanzante J R. 1996. Resistant, robust and non-parametric techniques for the analysis of climate data:Theory and examples, including applications to historical radiosonde station data[J]. International Journal of Climatology, 16(11):1197-1226, doi:10.1002/(SICI)1097-0088(199611)16:11<1197:AID-JOC89>3.0.CO;2-L.
    [18] 李军, 张友鹏. 2011. 基于高斯过程的混沌时间序列单步与多步预测[J]. 物理学报, 60(7):070513. http://www.cnki.com.cn/Article/CJFDTOTAL-WLXB201107021.htm

    Li Jun, Zhang Youpeng. 2011. Single-step and multiple-step prediction of chaotic time series using Gaussian process model[J]. Acta Physica Sinica, 60(7):070513. http://www.cnki.com.cn/Article/CJFDTOTAL-WLXB201107021.htm
    [19] 李永, 张美玲, 张军, 等. 2013. 单站实时地面观测数据质量控制方法的试验[J]. 北京农业, 9:146-147. http://www.cnki.com.cn/Article/CJFDTOTAL-BJNY201309114.htm

    Li Yong, Zhang Meiling, Zhang Jun, et al. 2013. Test of quality control method of single-station and real-time surface observations[J]. Beijing Agriculture (in Chinese), 9:146-147. http://www.cnki.com.cn/Article/CJFDTOTAL-BJNY201309114.htm
    [20] Lorenc A C. 1981. A global three-dimensional multivariate statistical interpolation scheme[J]. Mon. Wea. Rev., 109(4):701-721, doi:10.1175/1520-0493(1981)109<0701:AGTDMS>2.0.CO;2.
    [21] Lorenc A C, Hammon O. 1988. Objective quality control of observations using Bayesian methods-Theory, and a practical implementation[J]. Quart. J. Roy. Meteor. Soc., 114(480):515-543, doi:10.1002/qj. 49711448012.
    [22] Meek D W, Hatfield J L. 1994. Data quality checking for single station meteorological databases[J]. Agricultural and Forest Meteorology, 69(1-2):85-109, doi: 10.1016/0168-1923(94)90083-3.
    [23] Packard N H, Crutchfield J P, Farmer J D, et al. 1980. Geometry from a time series[J]. Physical Review Letters, 45(9):712-716, doi: 10.1103/PhysRevLett.45.712.
    [24] Reek T, Doty S R, Owen T W. 1992. A deterministic approach to the validation of historical daily temperature and precipitation data from the cooperative network[J]. Bull. Amer. Meteor. Soc., 73(6):753-762, doi:10.1175/1520-0477(1992)073<0753:ADATTV>2.0.CO;2.
    [25] 任芝花, 熊安元. 2007. 地面自动站观测资料三级质量控制业务系统的研制[J]. 气象, 33(1):19-24. doi: 10.3969/j.issn.1000-0526.2007.01.003

    Ren Zhihua, Xiong Anyuan. 2007.Operational system development on three-step quality control of observations from AWS[J]. Meteorological Monthly (in Chinese), 33(1):19-24, doi: 10.3969/j.issn.1000-0526.2007.01.003.
    [26] Rosenstein M T, Collins J J, De Luca C J. 1993. A practical method for calculating largest Lyapunov exponents from small data sets[J]. Physica D, 65(1-2):117-134, doi: 10.1016/0167-2789(93)90009-P.
    [27] 史丽萍, 孙宝元, 于浩洋. 2003. 切比雪夫多项式回归分析方法在测量数据处理中的应用[J]. 黑龙江大学自然科学学报, 20(2):78-81. doi: 10.3969/j.issn.1001-7011.2003.02.021

    Shi Liping, Sun Baoyuan, Yu Haoyang. 2003. An application of Tchebyshev polynomials recursive analysis method on metrical data processing[J]. Journal of Natural Science of Heilongjiang University (in Chinese), 20(2):78-81, doi: 10.3969/j.issn.1001-7011.2003.02.021.
    [28] Shi Y H, Eberhart R C. 1998. Parameter selection in particle swarm optimization[M]//Porto V W, Saravanan N, Waagen D, et al.Evolutionary Programming VII. Berlin Heidelberg:Springer, 591-600, doi: 10.1007/BFb0040810.
    [29] Takens F. 1981. Detecting strange attractors in turbulence[M]//Rand D, Young L S. Dynamical Systems and Turbulence, Warwick 1980:Lecture Notes in Mathematics. Berlin Heidelberg:Springer, 361-381, doi: 10.1007/BFb0091924.
    [30] 屠其璞, 王俊德, 丁裕国, 等. 1984. 气象应用概率统计学[M]. 北京:气象出版社, 183-186.

    Tu Qipu, Wang Junde, Ding Yuguo, et al. 1984.Probability Statistics of Meteorology Application (in Chinese)[M]. Beijing:China Meteorological Press, 183-186.
    [31] Wade C G. 1987. A quality control program for surface mesometeorological data[J]. J. Atmos. Oceanic Technol., 4(3):435-453, doi:10.1175/1520-0426(1987)004<0435:AQCPFS>2.0.CO;2.
    [32] 王海军, 刘莹. 2012. 综合一致性质量控制方法及其在气温中的应用[J]. 应用气象学报, 23(1):69-76. doi: 10.3969/j.issn.1001-7313.2012.01.008

    Wang Haijun, Liu Ying. 2012.Comprehensive consistency method of data quality controlling with its application to daily temperature[J]. Journal of Applied Meteorological Science (in Chinese), 23(1):69-76, doi:10.3969/j.issn.1001-7313.2012. 01.008.
    [33] 王海军, 杨志彪, 杨代才, 等. 2007. 自动气象站实时资料自动质量控制方法及其应用[J]. 气象, 33(10):102-109. doi: 10.3969/j.issn.1000-0526.2007.10.015

    Wang Haijun, Yang Zhibiao, Yang Daicai, et al. 2007. The method and application of automatic quality control for real time data from automatic weather stations[J]. Meteorological Monthly (in Chinese), 33(10):102-109, doi: 10.3969/j.issn.1000-0526.2007.10.015.
    [34] 肖方红, 阎桂荣, 韩宇航. 2005. 混沌时序相空间重构参数确定的信息论方法[J]. 物理学报, 54(2):550-556. doi: 10.3321/j.issn:1000-3290.2005.02.012

    Xiao Fanghong, Yan Guirong, Han Yuhang. 2005. Information theory approach to determine embedding parameters for phase space reconstruction of chaotic time series[J]. Acta Physica Sinica (in Chinese), 54(2):550-556, doi: 10.3321/j.issn:1000-3290.2005.02.012.
    [35] 熊安元. 2003. 北欧气象观测资料的质量控制[J]. 气象科技, 31(5):314-320. doi: 10.3969/j.issn.1671-6345.2003.05.013

    Xiong Anyuan. 2003. Quality control of meteorological observational data in Nordic countries[J]. Meteorological Science and Technology (in Chinese), 31(5):314-320, doi:10.3969/j.issn.1671-6345. 2003.05.013.
    [36] 杨荣华, 刘建华. 2011. 量子粒子群算法求解整数规划的方法[J]. 科学技术与工程, 11(33):8195-8198. doi: 10.3969/j.issn.1671-1815.2011.33.021

    Yang Ronghua, Liu Jianhua. 2011.Solving integer programming based on quantum particle swarm optimization[J]. Science Technology and Engineering (in Chinese), 11(33):8195-8198, doi: 10.3969/j.issn.1671-1815.2011.33.021.
    [37] 叶小岭, 周建华, 熊雄. 2014. 一种基于GEP的地面气温观测资料的质量控制方法[J]. 热带气象学报, 30(6):1196-1200. doi: 10.3969/j.issn.1004-4965.2014.06.021

    Ye Xiaoling, Zhou Jianhua, Xiong Xiong. 2014. A GEP-based method for quality control of surface temperature observations[J]. Journal of Tropical Meteorology (in Chinese), 30(6):1196-1200, doi: 10.3969/j.issn.1004-4965.2014.06.021.
    [38] 尹嫦姣, 江志红, 吴息, 等. 2010. 空间差值检验方法在地面气象资料质量控制中的应用[J]. 气候与环境研究, 15(3):229-236. doi: 10.3878/j.issn.1006-9585.2010.03.02

    Yin Changjiao, Jiang Zhihong, Wu Xi, et al. 2010. A research on the application of spatial difference method in quality control of surface meteorological data[J]. Climatic and Environmental Research (in Chinese), 15(3):229-236, doi: 10.3878/j.issn.1006-9585.2010.03.02.
    [39] 赵天保, 艾丽坤, 冯锦明. 2004.NCEP再分析资料和中国站点观测资料的分析与比较[J]. 气候与环境研究, 9(2):278-294. doi: 10.3969/j.issn.1006-9585.2004.02.005

    Zhao Tianbao, Ai Likun, Feng Jinming. 2004. An intercomparison between NCEP reanalysis and observed data over China[J]. Climatic and Environmental Research (in Chinese), 9(2):278-294, doi:10.3969/j.issn.1006-9585. 2004.02.005.
    [40] 赵天保, 符淙斌. 2006. 中国区域ERA-40, NCEP-2再分析资料与观测资料的初步比较与分析[J]. 气候与环境研究, 11(1):14-32. doi: 10.3969/j.issn.1006-9585.2006.01.002

    Zhao Tianbao, Fu Congbin. 2006. Preliminary comparison and analysis between ERA-40, NCEP-2 reanalysis and observations over China[J]. Climatic and Environmental Research (in Chinese), 11(1):14-32, doi: 10.3969/j.issn.1006-9585.2006.01.002.
    [41] 周青, 张乐坚, 李峰, 等. 2015. 自动站实时数据质量分析及质控算法改进[J]. 气象科技, 43(5):814-822. doi: 10.3969/j.issn.1671-6345.2015.05.007

    Zhou Qing, Zhang Lejian, Li Feng, et al. 2015. Quality analysis of real-time AWS data and algorithm improvement of quality control[J]. Meteorological Science and Technology (in Chinese), 43(5):814-822, doi:10.3969/j.issn.1671-6345. 2015.05.007.
  • 加载中
图(7) / 表(3)
计量
  • 文章访问数:  1532
  • HTML全文浏览量:  24
  • PDF下载量:  2812
  • 被引次数: 0
出版历程
  • 收稿日期:  2016-01-13
  • 网络出版日期:  2016-07-06
  • 刊出日期:  2017-02-01

目录

    /

    返回文章
    返回