中气旋与强对流风暴相关参数的演变关系
|
摘要
强对流天气对社会公众的危害极大,强对流风暴内部的中气旋通常与灾害天气相关.为了预测强对流天气的发生时间,研究了中气旋参数与强对流风暴参数的相关关系.针对强对流风暴参数与中气旋参数之间的变化规律,引入了时间序列关联规则的挖掘方法.收集了天津塘沽雷达站的24个包含中气旋的强对流风暴案例,使用强对流风暴参数和中气旋参数的时间序列来描述案例集合,并使用典型变量分析算法将归一化的时间序列数据降维,考虑到序列间的变化趋势和局部关键点的影响,提取单调性特征和局部极值点特征对时间序列符号化表示,将同一时间段内的符号组成事务集,使用Eclat算法发现频繁项集;并且类比于学习器的性能度量,提出评估两类数据变化规律的正比率与反比率概念.实验结果显示,强对流风暴参数与中气旋参数之间存在显著的相关性,其第1对典型变量的相关系数大于0.5;在支持度为0.05~0.30下,Eclat算法的运行时间小于0.7 ms,内存占用小于121 KB,均优于Apriori算法和FP-growth算法;强对流风暴参数与中气旋参数之间的正比率大于0.7,直观地反映出强对流风暴参数与中气旋参数之间变化趋势的相似程度,且再次验证了频繁项集的正确性.该研究为利用中气旋数据预测强对流天气的变化趋势与发生时间提供了理论依据.
Severe convective weathers are extremely harmful to the public. The mesocyclone inside a severe convective storm is usually related to disaster weathers. To predict the occurrence time of severe convective weathers,the correlations between the parameters of mesocyclones and those of severe convective storms were studied. Moreover,a method for mining time series association rules was introduced in order to evaluate the variation law between severe convective storm parameters and mesocyclone parameters. Twenty-four severe convective storm cases involving mesocyclones in the Tianjin Tanggu radar station were collected. The case set was described by a time series of severe convective storm parameters and mesocyclone parameters,and a variable analysis algorithm was used to reduce the normalized time series data. Considering the trend of the variation between sequences and the influence of local key points,the monotonic features and local extremum features were extracted to symbolize the time series. The symbols in the same period were composed into a transaction set,and the frequent itemsets were found using the Eclat algorithm. Furthermore,a performance metrics method that reflects the similarity of the trends between the two sets of time series data according to positive ratio and inverse ratio is proposed to evaluate the variation law of two sets of data. Experimental results show that there is a significant correlation between the severe convective storm parameters and the mesocyclone parameters. The correlation coefficient of the first pair of typical variables is higher than 0.5.Under the support degree of 0.05—0.30,the running time of the Eclat algorithm is less than 0.7 ms,and the memory occupancy is less than 121 KB,which are better than those of both Apriori algorithm and FP-growth algorithm;the positive ratio between the severe convective storm parameters and the mesocyclone parameters is greater than 0.7,which reflects the similarity degree between the severe convective storm and the mesocyclone parameters and verifies the correctness of the frequent itemsets. This study provides a theoretical basis for predicting the trend and occurrence time of severe convective weathers using mesocyclone data.
Severe convective weathers are extremely harmful to the public. The mesocyclone inside a severe convective storm is usually related to disaster weathers. To predict the occurrence time of severe convective weathers,the correlations between the parameters of mesocyclones and those of severe convective storms were studied. Moreover,a method for mining time series association rules was introduced in order to evaluate the variation law between severe convective storm parameters and mesocyclone parameters. Twenty-four severe convective storm cases involving mesocyclones in the Tianjin Tanggu radar station were collected. The case set was described by a time series of severe convective storm parameters and mesocyclone parameters,and a variable analysis algorithm was used to reduce the normalized time series data. Considering the trend of the variation between sequences and the influence of local key points,the monotonic features and local extremum features were extracted to symbolize the time series. The symbols in the same period were composed into a transaction set,and the frequent itemsets were found using the Eclat algorithm. Furthermore,a performance metrics method that reflects the similarity of the trends between the two sets of time series data according to positive ratio and inverse ratio is proposed to evaluate the variation law of two sets of data. Experimental results show that there is a significant correlation between the severe convective storm parameters and the mesocyclone parameters. The correlation coefficient of the first pair of typical variables is higher than 0.5.Under the support degree of 0.05—0.30,the running time of the Eclat algorithm is less than 0.7 ms,and the memory occupancy is less than 121 KB,which are better than those of both Apriori algorithm and FP-growth algorithm;the positive ratio between the severe convective storm parameters and the mesocyclone parameters is greater than 0.7,which reflects the similarity degree between the severe convective storm and the mesocyclone parameters and verifies the correctness of the frequent itemsets. This study provides a theoretical basis for predicting the trend and occurrence time of severe convective weathers using mesocyclone data.
引文
[1]程浩.中尺度气旋识别与三维重建[D].天津:天津大学电气与自动化学院,2007.Cheng Hao.Mesoscale Cyclone Identification and 3DReconstruction[D].Tianjin:School of Electrical and Automation,Tianjin University,2007(in Chinese).
[2]方翀,郑媛媛.新一代天气雷达中气旋产品特征值统计和个例分析[J].气象,2007,33(11):16-20.Fang Chong,Zheng Yuanyuan.The analysis of mesocyclone product from the Doppler weather radar[J].Meteorological Monthly,2007,33(11):16-20(in Chinese).
[3]吴芳芳,俞小鼎,张志刚,等.对流风暴内中气旋特征与强烈天气[J].气象,2012,38(11):1330-1338.Wu Fangfang,Yu Xiaoding,Zhang Zhigang,et al.The characteristics of the mesocyclone and severe weather associated with convective storms[J].Meteorological Monthly,2012,38(11):1330-1338(in Chinese).
[4]郑艳,俞小鼎,任福民,等.海南一次超级单体引发的强烈龙卷过程观测分析[J].气象,2017,43(6):675-685.Zheng Yan,Yu Xiaoding,Ren Fumin,et al.Analysis on a severe tornado process in Hainan triggered by supercell[J].Meteorological Monthly,2017,43(6):675-685(in Chinese).
[5]高晓梅,孙雪峰,秦瑜蓬,等.山东一次强对流天气的环境条件和对流风暴特征[J].干旱气象,2018,36(3):447-455.Gao Xiaomei,Sun Xuefeng,Qin Yupeng,et al.Environmental conditions and convective storm features of a severe convective weather process in Shandong Province[J].Journal of Arid Meteorology,2018,36(3):447-455(in Chinese).
[6]张艳.基于雷达图像的中气旋的识别及超折射回波的滤除[D].天津:天津大学电气与自动化学院,2009.Zhang Yan.Identification of Mesocyclones Based on Radar Images and Filtering out of Super Reflection Echoes[D].Tianjin:School of Electrical and Automation,Tianjin University,2009(in Chinese).
[7]Agrawal R,Srikant R.Fast algorithms for mining association rules in large databases[J].Computer Engineering&Applications,2000,15(6):619-624.
[8]Han J,Pei J,Yin Y.Mining frequent patterns without candidate generation[J].ACM SIGMOD Record,2000,29(2):1-12.
[9]Zaki M J.Scalable algorithms for association mining[J].IEEE Trans on Knowledge and Data Engineering,2000,12(3):372-390.
[10]Das G,Lin K I,Mannila H,et al.Rule discovery from time series[C]//Proceedings of the 4th International Conference on Knowledge Discovery and Data Mining.Mcnlo Park,USA,1998:16-22.
[11]Lin J,Keogh E,Lonardi S,et al.A symbolic representation of time series,with implications for streaming algorithms[C]//ACM SIGMOD Workshop on Research Issues in Data Mining and Knowledge Discovery.New York,USA,2003:2-11.
[12]Xue R,Zhang T,Chen D,et al.Sensor time series association rule discovery based on modified discretization method[C]//IEEE International Conference on Computer Communication and the Internet.San Francisco,CA,USA,2016:196-202.
[13]Guo Y,Ding X,Liu C,et al.Sufficient canonical correlation analysis[J].IEEE Transactions on Image Processing,2016,25(6):2610-2619.
[14]Hotelling H.Relations between two sets of variates[J].Biometrika,1936,28:321-377.
[15]曾海泉.时间序列挖掘与相似性查找技术研究[D].上海:复旦大学信息科学与工程学院,2003.Zeng Haiquan.Research on Mining and Similarity Searching in Time Series Data Base[D].Shanghai:School of Information Science and Technology,Fudan University,2003(in Chinese).
[16]Aggarwal C C,Sun Z,Yu P S.Fast Algorithms for online generation of profile association rules[J].IEEETrans on Knowledge and Data Engineering,2015,14(5):1017-1028.
[17]陈培恩.关联规则Eclat算法改进研究[D].重庆:重庆大学计算机学院,2010.Chen Pei’en.Improving Research on Association Rules Eclat Algorithm[D].Chongqing:School of Computer Science,Chongqing University,2010(in Chinese).
[18]Zaki M J.Scalable algorithms for association mining[J].IEEE Trans on Knowledge and Data Engineering,2000,12(3):372-390.
[2]方翀,郑媛媛.新一代天气雷达中气旋产品特征值统计和个例分析[J].气象,2007,33(11):16-20.Fang Chong,Zheng Yuanyuan.The analysis of mesocyclone product from the Doppler weather radar[J].Meteorological Monthly,2007,33(11):16-20(in Chinese).
[3]吴芳芳,俞小鼎,张志刚,等.对流风暴内中气旋特征与强烈天气[J].气象,2012,38(11):1330-1338.Wu Fangfang,Yu Xiaoding,Zhang Zhigang,et al.The characteristics of the mesocyclone and severe weather associated with convective storms[J].Meteorological Monthly,2012,38(11):1330-1338(in Chinese).
[4]郑艳,俞小鼎,任福民,等.海南一次超级单体引发的强烈龙卷过程观测分析[J].气象,2017,43(6):675-685.Zheng Yan,Yu Xiaoding,Ren Fumin,et al.Analysis on a severe tornado process in Hainan triggered by supercell[J].Meteorological Monthly,2017,43(6):675-685(in Chinese).
[5]高晓梅,孙雪峰,秦瑜蓬,等.山东一次强对流天气的环境条件和对流风暴特征[J].干旱气象,2018,36(3):447-455.Gao Xiaomei,Sun Xuefeng,Qin Yupeng,et al.Environmental conditions and convective storm features of a severe convective weather process in Shandong Province[J].Journal of Arid Meteorology,2018,36(3):447-455(in Chinese).
[6]张艳.基于雷达图像的中气旋的识别及超折射回波的滤除[D].天津:天津大学电气与自动化学院,2009.Zhang Yan.Identification of Mesocyclones Based on Radar Images and Filtering out of Super Reflection Echoes[D].Tianjin:School of Electrical and Automation,Tianjin University,2009(in Chinese).
[7]Agrawal R,Srikant R.Fast algorithms for mining association rules in large databases[J].Computer Engineering&Applications,2000,15(6):619-624.
[8]Han J,Pei J,Yin Y.Mining frequent patterns without candidate generation[J].ACM SIGMOD Record,2000,29(2):1-12.
[9]Zaki M J.Scalable algorithms for association mining[J].IEEE Trans on Knowledge and Data Engineering,2000,12(3):372-390.
[10]Das G,Lin K I,Mannila H,et al.Rule discovery from time series[C]//Proceedings of the 4th International Conference on Knowledge Discovery and Data Mining.Mcnlo Park,USA,1998:16-22.
[11]Lin J,Keogh E,Lonardi S,et al.A symbolic representation of time series,with implications for streaming algorithms[C]//ACM SIGMOD Workshop on Research Issues in Data Mining and Knowledge Discovery.New York,USA,2003:2-11.
[12]Xue R,Zhang T,Chen D,et al.Sensor time series association rule discovery based on modified discretization method[C]//IEEE International Conference on Computer Communication and the Internet.San Francisco,CA,USA,2016:196-202.
[13]Guo Y,Ding X,Liu C,et al.Sufficient canonical correlation analysis[J].IEEE Transactions on Image Processing,2016,25(6):2610-2619.
[14]Hotelling H.Relations between two sets of variates[J].Biometrika,1936,28:321-377.
[15]曾海泉.时间序列挖掘与相似性查找技术研究[D].上海:复旦大学信息科学与工程学院,2003.Zeng Haiquan.Research on Mining and Similarity Searching in Time Series Data Base[D].Shanghai:School of Information Science and Technology,Fudan University,2003(in Chinese).
[16]Aggarwal C C,Sun Z,Yu P S.Fast Algorithms for online generation of profile association rules[J].IEEETrans on Knowledge and Data Engineering,2015,14(5):1017-1028.
[17]陈培恩.关联规则Eclat算法改进研究[D].重庆:重庆大学计算机学院,2010.Chen Pei’en.Improving Research on Association Rules Eclat Algorithm[D].Chongqing:School of Computer Science,Chongqing University,2010(in Chinese).
[18]Zaki M J.Scalable algorithms for association mining[J].IEEE Trans on Knowledge and Data Engineering,2000,12(3):372-390.