A novel method for processing multivariate data of marine remote sensing

35th International Symposium on Remote Sensing of Environment

地球科学;生态环境科学

Ying, Chenlu^1,2 ; Dong, Qing^1 ; Xue, Cunjin^1 ; Wen, Xiaoyang^1

Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences, No.9 Dengzhuang South Road Haidian District, Beijing, 100094, China^1

University of Chinese Academy of Sciences, No.19A Yuquan Road, Beijing, 100049, China^2

关键词: Canonical correlation analysis;    Canonical correlations;    Correlation coefficient;    Linear correlation;    Marine remote sensing;    Maximum correlations;    Multiple correlation coefficients;    Ocean remote sensing;   

Others  :  https://iopscience.iop.org/article/10.1088/1755-1315/17/1/012209/pdf
DOI  :  10.1088/1755-1315/17/1/012209

学科分类：环境科学（综合）

来源: IOP

【 摘 要 】

According to current situations that large-scale, quantitative conjoint analysis methods of multivariate marine data are restricted to pairwise analysis, a novel method for processing multivariate data is put forward, applied to exploring the correlation in three marine factors in spatial distribution and time period. Based on canonical correlation analysis, this method takes the multiple correlation coefficient as a standard to measure the correlation of three factors, separates the linear correlation mode according to the principle of maximum correlation. This method is applied to a 144-month ocean remote sensing data in North Pacific, the first canonical correlation mode indicates sea surface wind over 0-10°N, 160°E-180° E, precipitation over 0°N-5°N and SST over 0°-15°N, 170°E-150°W have the most significant correlation, and the correlation coefficients are greater than 0.8. Anomalies in these regions are similar to anomalies ENSO caused. This method can reduce calculation times from three to one, and avoid ignoring the influence of the third ocean factor in present conjoint analysis, which makes results of this method closer to the ocean phenomenon.