Reliable dispersion measurement between two seismic stations is an essential basis of surface wave imaging. Noise source directivity has become an inescapable obstacle and a main concern for passive seismic survey: It basically breaks the principle of Green’s function retrieval in travel-time tomography; moreover, the azimuthal effect of heterogeneous ambient noise sources will inherently cause different levels of early arrival on cross-correlation functions, and the apparent velocity of surface waves can be overestimated by either multichannel slant stackings or interstation frequency–time analysis. Waveforms intrinsically contain the features of travel-time, energy and asymmetry in cross-correlation functions, and in return, they can be mapped into the causative noise sources and medium structures. Based on the theoretical framework of full waveform ambient noise inversion, we proposed a method to jointly invert noise source distributions and the corresponding unbiased surface wave velocities. The coupled dependencies of source distributions and path velocities in waveform misfit function show necessity of source–structure joint inversion. The decoupling strategy of partial derivatives is approved by the synthetic tests. Field experiments in the Hangzhou urban area further reveal the practicability of the theory. The inverted noise source models are comparable with the in situ noise distributions in urban environment, and the delineated surface wave velocities have been verified by local borehole datasets. Finally, we concluded that the developed waveform joint imaging algorithm can well relieve the dilemma of source induced velocity uncertainties.

两个地震台站之间可靠的频散测量是表面波成像的重要基础。噪声源方向性已成为被动地震测量的一个不可避免的障碍和主要关注点：它基本上打破了走时层析成像中格林函数反演的原理；此外，异质环境噪声源的方位角效应会固有地导致互相关函数上不同程度的提前到达，并且多通道倾斜叠加或站间频率时间分析可能会高估表面波的视速度。波形本质上包含互相关函数中的走时、能量和不对称性特征，作为回报，它们可以映射到成因噪声源和介质结构中。基于全波形环境噪声反演的理论框架，我们提出了一种联合反演噪声源分布和相应无偏表面波速度的方法。波形失配函数中震源分布和路径速度的耦合依赖性表明震源-结构联合反演的必要性。偏导数的解耦策略得到了合成测试的认可。在杭州市区的现场实验进一步揭示了该理论的实用性。反相噪声源模型与城市环境中的原位噪声分布具有可比性，所描绘的表面波速度已通过当地钻孔数据集进行验证。最后，