To make full use of seismograms to put tight constraints on the structure of subsurface and earthquake sources has always been the research focus in seismology. With increasing computational power, full waveform based seismic tomography has been applied in some regions with promising results. However, the heavy demand for computational resources and strong nonlinearity still prohibit its wide applications. Additionally, most applications of full waveform tomography at regional or global scales can only fit relatively long-period waveforms; the highest frequency of waveform fitting in full waveform tomography is approximately 0.5 Hz on regional scales and even lower on global scales. An alternative way to take advantage of more information on seismograms is the joint inversion of body and surface waves. Instead of fitting low-frequency waveforms, as in full waveform tomography, the joint inversion method uses high-frequency body-wave arrival times and surface-wave dispersion measurements. The forward problem in joint inversion only involves ray tracing or solving the Eikonal equation numerically. Therefore, it is less demanding in terms of computational resources. Compared to separate inversion using either body or surface wave data, joint inversion can provide a unified VP
model, and thus more reasonable VP
ratio model, by taking advantage of the complementary strength of both data sets. These models could impose tighter constraints on lithology, porosity, and partial melting. Moreover, machine learning-based techniques to detect earthquakes and pick arrivals have obtained many high-frequency arrival times on regional scales with dense deployments, which could be used in joint inversion to improve regional wavespeed models in the crust and upper mantle. The improved models may benefit other seismological studies and provide better understanding of regional tectonics. In this paper, we review some widely used seismic tomography methods for constructing regional models, introduce the basics of joint inversion and its application in southwest China, and discuss potential improvements.