Abstract: Source parameters such as focal mechanism solutions and depth are of significant value for studying regional geological structures, earthquake rupture processes, and seismogenic mechanisms. Currently, numerous methods exist for determining these source parameters, including P-wave first motion polarity inversion, W-phase inversion, and waveform fitting for focal mechanism solutions, as well as surface wave spectroscopy and depth phase methods for determining earthquake depth. Notably, the cut-and-paste （CAP） waveform fitting algorithm, which separates body and surface waves for inversion, utilizes depth phases （e.g., sPmP, sPg, sPn） and amplitude information to obtain focal mechanism solutions and focal depths, and is widely employed for regional seismic source parameter inversion. However, complex three-dimensional structures can alter waveform shape and amplitude, raising concerns among scholars about their impact on source parameter determination. Zhu and Zhou utilized three-dimensional Green’s function inversion for the 2013 Lushan M_S7.0 earthquake, finding that when the three-dimensional velocity model is reliable, the effectiveness of waveform fitting is significantly enhanced. Nonetheless, the computational cost and storage space required for synthesizing seismic waveforms using three-dimensional velocity structures are considerably high, making the process challenging and resource-intensive. Therefore, many scholars still employ one-dimensional velocity models for source parameter inversion. Given the current scarcity of research on applying three-dimensional velocity structure inversion for source parameter estimation, further quantitative studies on the impact of three-dimensional velocity models on source parameter inversion are warranted. This study focuses on the Luding area in Sichuan, located at the boundary between the Sichuan-Yunnan block and the Bayan Har block on the eastern edge of the Qinghai-Xizang Plateau. The area features significant differences in velocity structures on both sides, multiple active faults, strong crustal activity, numerous seismic events, and a dense station distribution, providing rich data support for studying the impact of complex three-dimensional structures on source parameters. This study utilizes the locations and mechanism solution information of 24 earthquakes with M_S≥3.5 in the Luding area from 2009 to 2019 as inputs, incorporates the impact of three-dimensional velocity structures, and synthesizes seismic waveforms using the spectral element method. The three-dimensional synthesized waveforms are then treated as observed waveforms, and various one-dimensional velocity models are tested to invert source parameters, thereby assessing the impact of three-dimensional structures on commonly used one-dimensional waveform fitting methods for obtaining source parameters. The results indicate that within the 0.02 to 0.1 Hz filtering band, the complexity of the three-dimensional velocity structure in the region has a minor impact on the inversion of source parameters for moderate and small earthquakes with magnitudes ranging from M_S3.5 to M_S5.0. The focal depth errors under all four models remain within 2 km, the magnitude errors are within 0.15, and the minimum rotation angles of the source mechanisms are all below 18°, with the inversion based on the average velocity model of the Longmenshan fault zone demonstrating the best performance. Additionally, testing results show that when studying areas with complex tectonics and topography, using a partitioned one-dimensional velocity model for inversion reduces the magnitude error to within 0.1 and the minimum rotation angle of the focal mechanism to within 12°, offering an enhanced reliability in source parameter acquisition compared with other one-dimensional velocity models. To assess the impact of inaccurate source locations on inversion outcomes, this study conducted inversion tests by shifting the earthquake locations from their true positions. The results indicate that the current precision of earthquake localization has a minor effect on the precise determination of source parameters, and the CAP method can reliably acquire source parameters for moderate and small earthquakes. Finally, using broadband waveform records from the Sichuan Seismic Network and based on the average velocity model of the Longmenshan fault zone, the study inverted the focal mechanism solutions for M_S≥3.5 earthquakes in the region from 2009 to 2019 using the CAP method. The inversion results demonstrate good waveform fitting, with minimal differences in parameters compared with published focal mechanism catalogs, further corroborating the reliability of the inversion outcomes. Combined with the precise localization results of the aftershocks of the Luding M_S6.8 earthquake occurred on September 5, 2022, the study analyzed the seismogenic tectonics and geometric structural characteristics in the Luding area. It was found that the inversion mechanism solutions for most events on the eastern side of the Xianshuihe fault zone exhibit sinistral strike-slip characteristics, with a NW-SE orientation, consistent with the overall strike of the Xianshuihe fault zone. Earthquake events near the Moxi fault show a regional distribution, with seismic events in the southern and northern segments exhibiting sinistral strike-slip fault characteristics, matching the characteristics of the Moxi fault. Some events in the western area of the middle segment of the Moxi fault exhibit normal fault characteristics. Combined with the research by Feng et al, it was found that there were a few oblique-slip normal fault events in the 2016 M_L4.2 Luding earthquake swarm, suggesting the presence of multiple unidentified normal fault branches on the western side of the middle segment of the Moxi fault. These results furnish a crucial scientific foundation for the assessment of tectonic activity and seismic hazard risk in the region.