时间序列分析在海表温度定量研究中的应用

发布时间：2018-05-25 05:19

本文选题：海表温度 + 时间序列分析　；参考：《杭州电子科技大学》2016年硕士论文

【摘要】：海表温度是监测海洋现象的重要参量,对海洋生态系统有很大的影响和作用,具有重要的研究价值,其在海洋动力学、海气相互作用、渔业经济研究和污染检测等方面有广泛应用。此前国内外关于海表温度的许多研究都基于遥感反演、遥感数据重构,对其并没有作进一步的研究。海表温度的预报也多以经验预报方法、数值预报方法和统计预报方法为主,这些预报方法精度有限。由于海表温度受各种因素的影响,使得海表温度时间序列呈现出明显的季节性变化特征。本文用时间序列分析的方法,研究包括东海、杭州湾、台湾海峡和南海在内的中国近海的海表温度温度及其预报工作。研究内容主要包括下面三个方面:一、海表温度时间序列的预处理。首先是海表温度时间序列的聚类分析,将相似度高的样本聚为一类,把每个研究区域上的576个样本划分为两个类,以样本点多的类为例来研究相关海域的海表温度。且本文提出了气候月的思想,对时间序列数据按气候月求月平均海表温度,提高了预测的精度。二、从时域分析的角度,对月平均海表温度时间序列数据,经过模型识别、模型估计和模型的诊断检验,建立SARIMA模型并作预测。三、从频域分析的角度,对月平均海表温度时间序列数据做谱分析,观察其周期图,建立相应的潜周期模型或混合潜周期模型并作预测。由拟合的SARIMA模型和潜周期模型分别预测2010年3至2011年2月这12个月的月平均海表温度,通过实际值和预测值的比较,发现这两种预报方法的预测精度都较高,可以为这些地区海表温度的研究提供参考。
[Abstract]:Sea surface temperature (SST) is an important parameter for monitoring ocean phenomena, which has great influence and effect on marine ecosystem, and has important research value in marine dynamics and air-sea interaction. Fishery economic research and pollution detection are widely used. Many previous studies on sea surface temperature are based on remote sensing inversion, remote sensing data reconstruction, and no further research has been done. The forecasting methods of sea surface temperature are mostly empirical, numerical and statistical, and the precision of these forecasting methods is limited. Because sea surface temperature is affected by various factors, the time series of sea surface temperature show obvious seasonal variation characteristics. In this paper, time series analysis is used to study the sea surface temperature (SST) and its prediction work in the East China Sea, Hangzhou Bay, Taiwan Strait and South China Sea, which include the East China Sea, Hangzhou Bay, Taiwan Strait and the South China Sea. The main contents include the following three aspects: first, the pretreatment of sea surface temperature time series. Firstly, the sea surface temperature (SST) time series is analyzed by clustering. The samples with high similarity are grouped into two categories, and the sea surface temperature (SST) in the relevant sea areas is studied by taking the multi-sample groups as an example. The idea of climate month is put forward in this paper, and the prediction accuracy is improved by calculating monthly mean sea surface temperature according to the time series data. Secondly, from the point of view of time domain analysis, the SARIMA model is established and predicted by model identification, model estimation and model diagnosis test for the monthly mean sea surface temperature time series data. Thirdly, from the point of view of frequency domain analysis, the spectral analysis of monthly mean sea surface temperature time series data is made, the period diagram is observed, and the corresponding latent period model or mixed latent period model is established and forecasted. Based on the fitting SARIMA model and the latent period model, the monthly mean sea surface temperature for 12 months from March 2010 to February 2011 is predicted respectively. By comparing the actual values with the predicted values, it is found that the two forecasting methods have higher prediction accuracy. It can provide reference for the study of sea surface temperature in these areas.
【学位授予单位】：杭州电子科技大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：P731.11

【参考文献】