The conventional energy flux density vector indicates the propagation direction of mixed P- and S-wave wavefields, which means when a wavefront of P-wave encounters a wavefront of S-wave with different propagation directions, the vectors cannot indicate both directions accurately. To avoid inaccuracies caused by superposition of P- and S-waves in a conventional energy flux density vector, P- and S-wave energy flux density vectors should be calculated separately. Because the conventional energy flux density vector is obtained by multiplying the stress tensor by the particle-velocity vector, the common way to calculate P- and S-wave energy flux density vectors is to decompose the stress tensor and particle-velocity vector into the P- and S-wave parts before multiplication. However, we have found that the P-wave still interfere with the S-wave energy flux density vector calculated by this method. Therefore, we have developed a new method to calculate P- and S-wave energy flux density vectors based on a set of new equations but not velocity-stress equations. First, we decompose elastic wavefield by the set of equations to obtain the P- and S-wave particle-velocity vectors, dilatation scalar, and rotation vector. Then, we calculate the P-wave energy flux density vector by multiplying the P-wave particle-velocity vector by dilatation scalar, and we calculate the S-wave energy flux density vector as a cross product of the S-wave particle-velocity vector and rotation vector. The vectors can indicate accurate propagation directions of P- and S-waves, respectively, without being interfered by the superposition of the two wave modes.

Elastic‐wave reverse‐time migration is an important imaging strategy because of its ability to image complex geologic structures and to efficiently extract elastic‐wave angle‐domain common‐image gathers (ADCIGs), which is an important task for amplitude‐versus‐angle (AVA) analysis and migration velocity analysis (MVA) in multicomponent exploration. Based on the characteristics of Poynting vectors in an elastic medium, we derive the wavefield‐separation imaging conditions for elastic reverse‐time migration (RTM) and establish the method of calculating the incident angle of source P‐wavefield using the propagation angle of source P‐wavefield and local dip angle of the reflector. Accurate ADCIGs can then be extracted based on the estimated incident angles with small computational cost. Two numerical examples show that the proposed method for extracting elastic‐wave ADCIGs can provide clear and accurate image and image gathers for elastic‐wave imaging tasks and therefore can provide reliable amplitude information for AVA and MVA. Copyright © 2016 John Wiley & Sons, Ltd.