Data Adaptive Analysis on Vertical Surface Deformation Derived from Daily ITSG-Grace2018 Model

Area of Science:

• Geodesy and Geophysics
• Earth Science
• Satellite Gravimetry

Background:

• Monthly gravity models are insufficient for detecting sub-monthly crustal displacement variations.
• Traditional least squares (LS) methods struggle with time-varying geophysical signals and missing data in geodetic time series.

Purpose of the Study:

• To analyze sub-monthly crustal vertical deformation in China using a high-temporal-resolution daily gravity model.
• To introduce and evaluate the data-adaptive Enhanced Harmonic Analysis (EHA) for geodetic time series analysis.
• To compare the performance of EHA against traditional LS methods for time-varying signal detection.

Main Methods:

• Utilized the daily ITSG-Grace2018 gravity model for 1.0° × 1.0° gridded vertical deformation in China.
• Introduced the Enhanced Harmonic Analysis (EHA) based on an Independent Point (IP) scheme.
• Performed comparative analysis between EHA and traditional Least Squares (LS) methods, assessing discrepancies and validating EHA's robustness with various noise types, IP numbers, and missing data scenarios.

Main Results:

• Daily gravity models are necessary for analyzing crustal displacement, with residuals up to 3.5 mm.
• EHA demonstrated significant advantages over LS in capturing time-varying signals, with a median discrepancy of 69.7% exceeding 10% across epochs.
• EHA effectively handles various noise types (white and color), is robust to the number of IPs, and performs well with continuous missing data without interpolation.

Conclusions:

• A higher temporal resolution gravity model is crucial for accurate crustal displacement analysis.
• EHA is a superior method for analyzing geophysical signals with time-varying amplitudes and missing data, outperforming traditional LS.
• The spatial distribution of discrepancies highlights regional characteristics and the limitations of assuming constant signal amplitudes in specific areas.