AUTHOR=Miao Peiyu , Xiao Genru , Wang Shengping , Zhang Keliang , Bai Buang , Guo Zeng 

TITLE=Effects of different seasonal fitting methods on the spatial distribution characteristics of common mode errors

JOURNAL=Frontiers in Earth Science

VOLUME=Volume 11 - 2023

YEAR=2023

URL=https://www.frontiersin.org/journals/earth-science/articles/10.3389/feart.2023.1176241

DOI=10.3389/feart.2023.1176241

ISSN=2296-6463

ABSTRACT=We investigated the effects of various seasonal fitting techniques on the spatial distribution of the common mode errors using the coordinate time series of the continuous GPS reference stations of CMONOC as an example. First, the seasonal term of coordinate time series was calculated using constant amplitude harmonic fitting (CAF), continuous wavelet transform   (CWT), and smoothing spline fitting (SPF). The seasonal term and linear trend were then removed to obtain the residual time series. In order to determine the common mode errors of residual time series, we apply PCA, finally. The results indicate that: (1) SPF is superior to CAF and CWT in its ability to fit short-term irregular seasonal signals. In comparison to CAF, N/E/U has RMSE values of SPF that are 25%, 20%, and 14.4% lower, respectively. SPF also has a higher coefficient of determination than CWT and CAF. The coefficient of determination in the U direction is larger than N and E directions; (2) Each order PC of the residual series fitted by SPF exhibits apparent spatial aggregation characteristics, with PC1 having a uniform spatial distribution and presenting a large positive response. The residual series obtained by CAF and CWT exhibits scattered spatial response distribution features in each order PC, nevertheless. Compared to N and E, U's spatial response distribution is distinct. From north to south, the spatial response of PC1 in U progressively diminishes. In addition to being much lower than that in other locations, the Sichuan-Yunnan region's spatial response value of PC1 and PC3 also exhibit a clear negative reaction. The RMSE value of the residual series after SPF is the least and the filtering effect is the best when comparing the spatial filtering effect based on the three fitting methods. We also compare the RMSE reduction ratio before and after spatial filtering, the results show that the RMSE reduction ratio before and after the residual series obtained by SPF is slightly larger than other methods.