Friday, August 27, 2010

Marine Multi-channel Seismic Processing (part-2)

(4) Pre-process

With the raw shot data with geometry laoded, we are looking forward to seeing changes happen to make the data better till the final image of cross-section. The first thing we do is to use Ormsby bandpass filter to remove the noise generated during acquisition (ProMAX module: Bandpass Filter). Remember we have done the main frequency range analysis when we get raw shot data by using Interactive Spectral Analysis module. Take it and use it in bandpass filter. You can see big difference after employing this filter to the raw shot. The second thing we do is to edit the traces, including to kill bad channels (ProMAX module: Trace Kill/Reverse) and to remove the spikes and brusts (ProMAX module: Spike and Noise Edit). Remember we have known the bad channels using Trace Display when we just get the raw shot data in hand. So input the information in Trace Kill to get rid of those bad ones. All right, after these two steps, we have been able to see the difference from raw data. It is much better, isn't it?

But not good enough. The third thing we do in the pre-process is deconvolution. With the help of deconvolution, we could enhance the primaries and suppress the multiples (ProMAX module: Spking and Predictive Decon). Here, we need to test some critical parameters of the deconvolution, to figure out which ones create best results. It takes time! Please be patient, and read the related books and papers to understand how the deconvolution works and how it could work better.

(5) Velocity Analysis

After the pre-process flow, we have got better-look data in hand. It is time for us to do velocity analysis (ProMAX module: Velocity Analysis), which will take a lot time to complete. So be patient enough to get this step done. First of all, we start with large CDP interval, for example, 5000 CDP interval in a section of 60000 CDPs. When we conduct the velocity analysis, remember to use the near-trace plot that we have made before so that we could recognize the main horizons, and keep the direct wave, primaries and multiples paths in mind, so as to distinguish the primaries from multiples. Try our best to keep veolocity analysis away from multiples, and we know it is not always easy honestly.

Technically, we have stack velocity and interval velocity during velocity analysis, while stack one is lower than the interval one. However, we try our best to keep both of them increase reasonably with increasing depth. Because the seismic velocities of different layers or horizons will increase with depth in common cases due to the increase of some physical attributes like density. And the main factor for quality control during velocity analysis is to see the flat horizons after applying NMO (Normal Moveout). That is to say, if we pick the accurate velocity for the certain horizon, we are able to see the straight or flat coherent event in the trace gather. Sometimes we have some obvious coherent event to apply NMO to make sure the correct velocity we pick, especially in the upper layers, but sometimes not, especially in the lower layers. So in this unlucky situation, we would like to use the semblance graph to find out the energy concentration hotspot, and together with the increasing velocity with increasing depth in mind, to pick predictive velocities.

Again, be carefull of multiples. Because they will show up with hotspots in the semblance which might get you confused in some points, but the distinct thing is that they just have the same or similar velocity all the way down with the increasing depth, i.e., the multiples' velocity function curve should be a nearly straight line from the top down to the bottom. Anyway, try our best to be away from multiples during the whole process of velocity analysis. As long as we build up the brute velocity model with the large CDP interval, we can conduct the so-called brute stack. When we want to see more details for the strutures or something interesting, we need to denser the CDP interval for velocity analysis, for example, go for 2500 CDP, 1000 CDP interval, or even smaller CDP interval for some specific areas to image the relative smaller struture. So it depends on where is the interesting place we want to look at, how much details we want to see and what geolocial question we want to answer.

To Be Continued, see you next week, part 3!

No comments:

Post a Comment