Gravity-Constrained Three-Dimensional Crustal Density Structure of the Sichuan–Yunnan Region Based on HMRF Model Integration and Bayesian Inversion

The Sichuan–Yunnan region is characterized by strong tectonic deformation, high seismicity, and frequent destructive earthquakes. Consequently, constructing accurate crustal physical models is essential for elucidating seismogenic mechanisms and tectonic processes. However, owing to inconsistencies in observational datasets, differences in data processing strategies, and limitations in computational approaches, crustal models proposed by different researchers for this region remain partially inconsistent.To address these issues, this study first derives a representative common model by integrating multiple existing geophysical models using a Hidden Markov Random Field (HMRF) framework. Based on this common model, a three-dimensional crustal density structure model is constructed through Bayesian inversion, incorporating constraints from Bouguer gravity anomaly.The results reveal pronounced spatial variations in density structure beneath major faults in the region. Distinct positive lateral density anomalies are identified beneath the Yuanmou, Qujing, Chenghai, Lijiang–Xiaojinhe, Xianshuihe, and Zhaotong–Ludian faults, whereas comparable anomalies are absent beneath several other fault zones. In addition, density distributions exhibit significant along-strike heterogeneity within individual faults, reflecting variations in deep crustal architecture. Notably, the Lijiang–Xiaojinhe fault displays the most prominent signatures of deep-seated material intrusion, while density anomalies associated with the Red River fault show a systematic deepening trend from east to west.