| 数据挖掘算法在河口羽状流数据分析中的应用 |
| |
| 引用本文： | 李昭颖,王厚杰.数据挖掘算法在河口羽状流数据分析中的应用[J].中国海洋大学学报(自然科学版),2021(3). |
| |
| 作者姓名： | 李昭颖 王厚杰 |
| |
| 作者单位： | 中国海洋大学海洋地球科学学院 |
| |
| 基金项目： | 国家杰出青年科学基金项目(41525021)资助。 |
| |
| 摘 要： | 羽状流的近场扩散角作为河口动力学的关键参数,对河流入海物质的输运扩散过程有重要影响,而在真实环境中由于受到复杂因素的影响,羽状流扩散角存在着显著的变化。数据挖掘技术尽管在地学数据的应用上尚处于探索阶段,但研究这一问题提供了一种有效途径。本文基于马格达莱纳河羽状流近场扩散角及周边环境数据,利用多重数据挖掘算法开展数据分析和评估,最终建立了羽状流扩散模型,对其影响因素和时间序列开展研究。结果表明,数据挖掘算法能够有效指示环境因素的贡献值而不受量纲、极值的影响,从而为进一步探究河口羽状流控制因素提供参考。受数据量和数据精度的影响,随机森林算法得到的变量分析结果与实际情况更为相似,数据集特征与数据量之间的博弈是造成模型结果差异的主要原因。根据数据挖掘算法结果,可以在马格达莱纳河河口建立羽状流扩散角模型,并分析其随环境参数和时间变化的趋势。模型结果显示,对马格达莱纳河而言,环境因素与羽状流扩散角均存在负相关关系,河流的指向方向为最主要影响因素,其次为风速,因此在开展进一步分析时可以针对环境因素添加不同权重以期获得更真实的结果。随时间增长,羽状流扩散角将进一步收紧,可能会引起河口的通道化。
|
| 关 键 词： | 数据挖掘 随机森林 数据分析 河口羽状流 |
Application of Data Mining Technique to the Study of River Plume Extension |
| |
| LI Zhao-Ying,WANG Hou-Jie.Application of Data Mining Technique to the Study of River Plume Extension[J].Periodical of Ocean University of China,2021(3). |
| |
| Authors: | LI Zhao-Ying WANG Hou-Jie |
| |
| Institution: | (College of Marine Geosciences, Ocean University of China, Qingdao 266100,China) |
| |
| Abstract: | River plume spreading angle is the key factor in river dynamic,which has important influence on the transport and diffusion process of rivers to the sea.However,plume spreading angle has significant variation in real environment due to the influence of complex factors.Thus,data mining technology provides an effective way of solving this problem though it is still in the exploratory stage of geoscience data application.Based on the Magdalena River plume near-field spreading angle and its surrounding environment data,multiple data mining algorithms are used for data analysis and evaluation,while finally a plume spreading model is established to study its influencing factors and time series.The results show that data mining algorithm can indicate the contribution value of parameters without being affected by the dimension and extreme value,providing a reference for further exploration of the control factors of the plume in the estuary.Affected by data volume and data accuracy,the variable analysis results obtained by the random forest algorithm are considered to be more similar to actual situation.The main reason for differences in algorithm result is caused by the competition between dataset feature and dataset amount.According to the results of data mining algorithms,a plume spreading angle model can be established at the estuary of Magdalena River,and its trend with environmental parameters and time can be analyzed.Model results show that there is negative correlation between the environmental factors and the plume spreading angle for Magdalena River,while discharge direction is the most important factor,followed by the wind speed,which can be used for further analysis in river sedimentation and accumulation to add different weights on environmental factors obtaining more realistic results.Over time,the plume spread angle will further tighten,which may cause channelization of the estuary. |
| |
| Keywords: | data mining random forest data analysis river plume |
| 本文献已被 CNKI 维普 等数据库收录！ |
|
