基于支持向量机的珠江河口咸潮上溯预测研究
Study on Prediction of Saltwater Intrusion Based on Support Vector Machine in the Pearl River Estuary

DOI: 10.12677/JWRR.2015.41002, PP. 8-16

Keywords: 珠江河口区，咸潮上溯，支持向量机
Pearl River Estuary, Saltwater Intrusion, Support Vector Machine

Full-Text   Cite this paper   Add to My Lib

Abstract:

受径流、潮汐、地形、风向等多重因素影响，珠江河口区咸潮上溯现象异常复杂。本文运用支持向量机回归(SVR)与分类(SVC)联合建模的方法，综合考虑上游径流和河口区潮汐因素，构建珠江河口区磨刀门水道咸潮上溯预测模型，并采用动态反馈机制分别开展了磨刀门水道日均咸度和超标历时预测。研究结果表明：基于支持向量机的咸潮上溯模型，具有模型简单精度高，样本容量要求少，自学习与联想记忆能力强的特点，能较好模拟预测多重因素影响的复杂河口区咸潮上溯现象；磨刀门水道上游三水马口前三日流量、河口区前三日三灶站的平均潮差过程与咸潮上溯具有较好时间相关性。
Effected by multiple factors such as upstream runoff, tide, topography, and wind, the saltwater in-trusion of Pearl River Estuary displays extreme complexity. In this paper, a hybrid method based on Support Vector Machine (SVR) and Classification (SVC) is used to build a daily average salinity prediction model in the Pearl River Estuary which has considered the variation of upstream runoff and tide. The result shows that the model for saltwater intrusion prediction has high accuracy and only requires a few samples. Furthermore, the model has a good ability of self-learning and asso-ciative memory which can well simulate the complex saltwater intrusions of this estuary. The findings of this study also show that the previous average three-day runoff from upstream and the previous three-day tidal range from outlet have a good time correlation with saltwater intrusion in this estuary.

References

[1]	ZHANG X.-F., DENG J.-Q. Affecting factors of salinity intrusion in Pearl River Estuary and sustainable utilization of water resources in Pearl River Delta. Sustainability in food and water. Netherlands: Springer, 2010: 11-17.
[2]	宋志尧, 茅丽华. 长江口盐水入侵研究[J]. 水资源保护, 2002, 3, 27-30. SONG Zhiyao, MAO Lihua. Saltwater intrusion at Yangtze River Estuary. Water Resources Protection, 2002, 3: 27-30. (in Chinese)
[3]	章文, 刘丙军, 辛彦博, 等. 珠江河口区盐度变化周期特征分析——以磨刀门水道为例[J]. 热带地理, 2013, 33(1): 28-33. ZHANG Wen, LIU Bingjun, XIN Yanbo, et al. Analysis of the characters of saltwater intrusion in Pearl River Estuary based on salinity data of Modaomen Waterway. Tropical Geography, 2013, 33(1): 28-33. (in Chinese)
[4]	闻平, 陈晓宏, 刘斌, 等. 磨刀门水道咸潮入侵及其变异分析[J]. 水文, 2007, 27(3): 65-67. WEN Ping, Chen Xiaohong, LIU Bin, et al. Analysis of tidal saltwater intrusion and its variation in Modaomen Channel. Journal of China Hydrology, 2007, 27(3): 65-67. (in Chinese)
[5]	欧素英, 杨青书, 雷亚平. 咸潮入侵理论预报模式的分析及其在西江三角洲的应用[J]. 热带海洋学报, 2010, 29(1): 32-41. OU Suying, YAMG Qingshu and LEI Yaping. Analysis of a model for saline water intrusion and its application in the West River Delta. Journal of Tropical Oceanography, 2010, 29(1): 32-41. (in Chinese)
[6]	SAVENIJE, H. H. G. Sa-linity and tides in alluvial estuaries. Amsterdam: Elsevier, 2005.
[7]	VAPNIK, V. The nature of statistical learning theory. New York: Springer-Verlag, 1995.
[8]	刘德地, 陈晓宏. 基于偏最小二乘回归与支持向量机耦合的咸潮预报模型[J]. 中山大学学报(自然科学版), 2007, 46(4): 89-92. LIU Dedi, CHEN Xiaohong. Model for prediction of saltwater intrusions based on coupling of support vector machine and partial least square method. Acta Scientiarum Naturalium Universitatis Sunyatseni, 2007, 46(4): 89-92. (in Chi-nese)
[9]	GONG, W., SHEN, J. The response of salt intrusion to changes in river discharge and tidal mixing during the dry season in the Modaomen Estuary, China. Continental Shelf Research, 2011, 31(7): 769-788.
[10]	路剑飞, 陈子燊. 珠江口磨刀门水道盐度多步预测研究[J]. 水文, 2010, 30(5): 69-74. LU Jianfei, CHEN Zishen. Salinity prediction at Modaomen Waterway in estuary of Pearl River. Journal of China Hy-drology, 2010, 30(5): 69-74. (in Chinese)
[11]	HSU, C. W., CHANG, C. C. and LIN, C. J. A practical guide to support vector classification. Technical Report, Department of Computer Science and Information Engineering, University of National Taiwan, Taipei, 2003: 1-12.
[12]	KOHAVI, R. A study of cross-validation and bootstrap for accuracy esti-mation and model selection. IJCAI, 1995, 14(2): 1137-1145.

Full-Text