Tracking the source direction of surface mass loads using vertical and horizontal displacements from satellite geodesy: A case study of the inter-annual fluctuations in the water level in the Great Lakes

Abstract

Climate oscillations with seasonal and longer periods drive surface water cycles and quasi-cycles at regional and global scales. Changes in terrestrial water storage produce responses in the Earth's gravitational field and crustal deformation. Here, we use techniques from the Global Positioning System (GPS) and the Gravity Recovery and Climate Experiment (GRACE) to reveal a normal pattern of inter-annual fluctuations in the water level in the North American Great Lakes (GL). The GRACE-estimated time series was in good agreement with in situ water level measurements from 2002 to 2018, especially in terms of phases. The amplitude of 3–4 inter-annual signals in water thickness was 4–6 cm, which is equivalent to a 50–75 km3 oscillation in surface water volume within the entire GL. The slightly larger annual and inter-annual fluctuation amplitudes estimated using GRACE data indicate that the aquifer system of the GL and its surroundings also contributes to seasonal mass changes. After 2013, water levels in the GL region rose abruptly, and the water mass increased by nearly 270 km3 until the end of 2018. We also used GPS- and GRACE-derived three-component displacements (vertical, northward, and eastward) to identify load patterns. GPS- and GRACE-estimated maximum probability source directions based on multi-channel singular spectrum analysis showed that most of the selected GPS sites point to the GL region, although the direction deviations of GPS results at a few sites are mainly caused by the combined effect of the local load and superposition of the distant GL. Our findings indicate that inter-annual displacement changes at different frequencies (1- to 8-year cycle) are primarily due to water volume fluctuations in the GL. The amplitudes estimated by GPS are greater than the GRACE-based and GL-modeled results at most stations, which reflects sensitivity differences between these geodetic solutions to the surface load, as hydrological processes in the local area around a station are difficult to identify using GRACE data with a resolution of ~300 km.