Acta Geodynamica et Geomaterialia

Title: COMPARATIVE ANALYSIS OF PCA AND ICA ON TREND ESTIMATION OF SEA-LEVEL CHANGE FROM TIDE GAUGE OBSERVATIONS
Authors: Wang Jie, He Xiaoxing, Hu Shunqiang, Sun Xiwen, Wang Wentao and Liu Huijuan
DOI: 10.13168/AGG.2024.0007
Journal: Acta Geodynamica et Geomaterialia, Vol. 21, No. 1 (213), Prague 2024
Full Text: PDF file (0.7 MB)
Keywords: Tide Gauge; Sea-Level Change; Trend Estimation; Stochastic Noise; PCA; ICA
Abstract: Sea-level rise directly caused by climate change is impacting coasts around the world and low-lying islands, requiring a continuous accurate monitoring. We analyze the sea-level data observed by 20 tide gauges located in the east coast of the United States of America (USA) over the period January 1972 to December 2021 by using an open-source toolbox SLR_APP. After mitigating noise using Principal Component Analysis (PCA) and Independent Component Analysis (ICA) method, we estimate the trend change and its uncertainty of sea-level considering the stochastic noise properties of the observations. The sea-level estimates and associated uncertainty are smaller than the raw observations after the noise reduction. Our results show that: the average values of the absolute trend change are 1.43 % and 0.78 %, and the mean trend uncertainty are reduced by 44.78 % and 21.26 % after PCA and ICA noise reduction, respectively. We conclude that PCA method performs better than ICA especially in reducing the associated trend uncertainty of the sea level change. Improving the sea-level rise estimation and prediction contribute globally to enhance public safety, in particular for for the coastal communities.