In recent years, the growing global scarcity of land resources has forced resource exploration shifted from the land to the sea. China has invested a lot of effort and financial resources in marine oil and gas exploration which concentrated in the deep waters of Qiongdongnan Basin, located in the northern part of South China Sea. According to the previous geophysical studies (Tao et al., 2005; Li, 2006; Duan and Zhang, 2006), it is known that the area is characterized by a complex geological structure, with a sudden increase in water depth over a wide range, which changes from a few hundred meters to a few kilometers. As a result, the dip angle suddenly becomes steeper and the sea floor undulates strongly. This kind of structure influences the propagation of seismic wave field energy greatly, thereby affecting the imaging quality of seismic data significantly. Most 2D seismic profiles from this area have the following problems: (1) the structure of the sunken floor is not clear; (2) the basal surface cannot be imaged properly; (3) the reflection is disordered; (4) the signal to noise ratio in the middle and deep layers is low; (5) there are strong multiple waves; (6) the side waves have developed (Zhu et al., 2008; Deng et al., 2010).

Previous researchers have introduced a lot of improvements in data acquisition and processing, including: expansion of the air gun capacity (or air gun combination), adjusting the cable length and sinking depth, and changing the direction of acquisition (Yang and Li, 2010; Dan et al., 2011); stratifying the velocity of seawater according to temperature and salt structure (Munk, 1974; Song et al., 2010; Han et al., 2012); improving energy compensation in the middle-deep layer and improving the signal to noise ratio technology using 2.5D geometric diffusion compensation factor (Han et al., 2013); and pre-stack depth migration processing for rugged seafloor (Chen and Ge, 2005; Chang et al., 2008). All these efforts have indeed improved the quality of seismic data acquired from steep slopes in deep waters, but there is still a large gap in the imaging quality when compared with data acquired from shallow waters or flat areas. In the land seismic data processing, the velocity models obtained from velocity analysis are fixed in the migration process. This means that a velocity model does not change in the process of migration imaging. However, for the processing of marine geophysical exploration data, it is paramount to take into account that seawater is affected by wind-waves, turbulence, etc., and that the topography of the seafloor may change at different times and in different degrees. Therefore, if we still use the fixed velocity model derived from the original velocity analysis for migration imaging processing, errors are likely to occur. These errors can weaken the middle-deep image field energy and also blur the migration profile.

To further understand other factors that affect the migration image field energy in the middle-deep layer and improve the imaging quality of the migration profile, it is important to consider the impact of seafloor topography over time on wave field recording and migration imaging. For data acquired in marine geophysical exploration, we assume that the data acquired on the seafloor Model A from Time T₁ to T_i is DATA1. As time goes by, at a Time T_i+1, the seafloor topography changed because of correlative factors, such as waves. We take the seafloor after the change in topography to be Nodel B and the seismic data acquired on Model B from Time T_i+1 to Time T_N as DATA2. However, during normal data processing, we cannot know the changes in velocity model at Time T_i+1. We will use the seafloor Model A to perform migration imaging based on the two datasets, DATA1 and DATA2. This study will focus on how the change in seafloor topography would influence the wave field and the migration imaging.

With regard to the influence of changes in the undulating seafloor on the wave field and migration image field, this study will establish two velocity models for analysis and discussion under the premise of ensuring that the stratum below the undulating seafloor is unchanged. (1) The velocity Model IA and the velocity Model IB when there is only a slight change in seafloor topography; (2) the velocity Model IIA and the velocity Model IIB when there is significant change in seafloor topography. The slight and significant change velocity models of the seafloor topography established here are mainly measured according to the difference between some acquired seismic wavefield records from South China Sea.

The research methods used in this study are as follows. We used finite difference method to generate synthetic seismic reflection dataset DATA1 on Model A from Time T₁ to T_i, at the Time T_i+1, seafloor topography had changed, and thus the velocity model changed to B. We generate synthetic seismic reflection data sets DATA1 on Model B from Time T_i+1 to T_N. Then we processed the synthetic datasets; we name the migration profile obtained by the wave field record DATA1 on Model A as P1, the migration profile obtained by the wave field record DATA1 and DATA2 on Model A as P2, and the migration profile obtained by the wave field record DATA2 on Model B as P3. We compared and analyzed the migration Profiles P1, P2 and P3 to evaluate how the seafloor topography changes will influence the migration image profiles. The results can also help to determine whether this method is effective enough to solve the problems of weaker middle and deep energy and fuzzy horizons in marine geophysical exploration data processing. The research workflow used here is shown in Fig. 1.

In the following, according to the research method and the processing flow, the influences on the wave field recording and the migration imaging will be analyzed in detail from the slight changes and significant changes in the undulating seabed.

In this paper, we try to find an explanation as to why migration imaging is blurred, and field energy is weaker in middle-deep layers when acquired the seismic reflection data from the undulating seafloor at certain times. We then analyzed and discussed the influence of both slight and significant changes on the seafloor topography on wave field records and migration imaging profiles. The following conclusions can be drawn from this study.

(1) For the velocity model with slight change, by comparing the wave field records acquired at the same point on original Models IA and IB, it can be seen that there is a slight difference at the point far offset and a slight influence on the energy level in the middle-deep layers.

(2) For the velocity model with significant change, by comparing the wave field records acquired at shot Points (50, 2), (250, 2), (2000, 2) and (2250, 2) on both Models IIA and IIB, it can be seen that there is an obvious difference in the distant trace and in the energy of the middle-deep layers.

(3) For the velocity model with slight change, whether seismic reflection data is acquired along the direction of the survey line or against the direction of the survey line, there is no difference between the profile of the single shot record on the original model and the model with changes. Therefore, the difference in the whole overlay profile is not significant.

(4) For the velocity model with significant change, whether seismic reflection data is acquired along the direction of the survey line or against the direction of the survey line, there is obvious difference between the profile of the single shot record on the original model and the model with changes. By comparing the results, it is clear that if the wrong model is used, the middle-deep layers will be blurred and severely deformed and influence the image field energy in the entire stack profile as a result.