Chongyang Shen

Three-dimensional gravity inversion technology involves inferring the underground density structure based on observed gravity anomaly data. In addition to gravity inversion based on physics-driven methods, deep learning, as a purely data-driven technique, is increasingly gaining attention in geophysical inversion problems. However, purely data-driven methods rely on the implicit relationships within the data during the inversion process, which results in a lack of clear physical significance. This study proposes a three-dimensional gravity inversion method that integrates physical equations with deep learning. Based on the U-Net architecture, the gravity forward equation is incorporated as a physical constraint term, and a composite loss function—comprising three-dimensional mean squared error, a depth-weighting function, and three-dimensional intersection-over-union loss—is constructed to enhance inversion accuracy. Numerical experiments indicate that this method outperforms traditional algorithms in terms of density recovery accuracy and boundary clarity. When applied to gravity anomaly data from the Tangshan earthquake region in China, this method successfully inverted the three-dimensional subsurface density structure, revealing a high-density anomaly beneath the seismic source area, which provides important evidence for understanding the regional earthquake generation mechanism.

The Barkam–Jiuzhaigou–Wuqi gravity profile extends across the Jiuzhaigou Ms7.0 earthquake (in 2017) zone and passes through several historical big earthquakes’ zones. We have obtained Bouguer gravity anomalies along the profile composed of 365 gravity observation stations with Global Positioning System (GPS) coordinates, analyzed the observed data and inverted subsurface density structure. The results show that the Moho depth has a big lateral variation from southwest to northeast, which shallows from 57 km to 43 km with maximum variation up to 14 km within 800 km. The most acute depth change of the Moho is in the boundary region between the Bayan Har block and West Qinling–Qilian block. According to our analysis, it is related to the eastward movement of the Bayan Har block. There are three main pieces of evidence that support it: (1) Density is higher in the east of the Bayan Har block and smaller in the west, which is the same as seismic activity; (2) Two thin low-density layers exist in the upper and middle crust of the Bayan Har block, which may promote inter-layer slip and the Jiuzhaigou Ms7.0 earthquake occurred in the boundary area of the two low-density layers, where the crustal density and Moho surface fluctuate sharply; (3) the GPS velocity field in the southwestern part gravity profile is significantly larger than that of the northeastern part, which is consistent with the density structure. Our studies also suggest that the large undulation of the Moho prevents the movement of the Bayan Har block, and strain is prone to accumulate here. The dynamic background analysis of the crust in this area indicates that the Moho surface uplifts in the West Qinling–Qilian block, which decelerates the eastern migration of material on the Qinghai–Tibet Plateau, and leads to the weak tectonic activity of the north part of the Bayan Har block.

The pattern evolution and dynamic mechanism of the dynamic changes of regional gravity fields occurring before and after the Wenchuan Ms8.0 earthquake are analyzed, based on five epochs of 1998–2007 mobile gravity data from the middle-south section of the north–south seismic belt, and two epochs of field research data collected after the 2008 Wenchuan earthquake in combination with GPS data, leveling observations, and geotectonic environment data. The regional dynamic gravity changes demonstrate the effects of the eastward flow of solid matter in the Qinghai-Tibetan plateau and the preparation of the 2008 Wenchuan earthquake (2–10yr). The two most meaningful gravity indicators of the Wenchuan earthquake preparation are the positive (increasing) gravity changes occurring over many years in the southwest epicenter and the large-scale gradient zone of gravity variation, with the cumulative difference between the two sides of the gradient zone of gravity exceeding 200 μGal. The positive gravity changes may facilitate a constant energy accumulation and the gradient belt may support seismic shear breakage. Overall, the gravity changes associated with the earthquake preparation indicate a pattern of accelerating increase-decelerating increase-earthquake occurrence. The Songpan-Ganzi block generally displays a negative gravity change, providing evidence for a local upwarping of the deep crust-mantle and an interior expansion of the deep crust attributable to high temperatures. The viewpoint is consistent with the dilatant mechanism for earthquake preparation