Friday, August 13, 2010

Quiz: How many iPods does it take to store reflection seismic data?


Hi all, it's another day on the Langseth, and we are still recording multichannel seismic (MCS) data. This is reflection data collected by hydrophone's trailing behind the ship. Yesterday, Sam was kind enough to give a window into the actual deployment of the streamers (housing the hydrophone) and the depth control devices, otherwise known as birds. Today I want to give the reader a feel for the amount of data memory involved in the actual seismic survey. This is the juncture where geophysics meets computer science. I'll keep it brief.

Let's start with the set up of the reflection experiment. Behind the ship we are currently towing a streamer cable that is approximately 6 km long. On this streamer cable there are 468 channels each recording data every 2 milliseconds. It might not be immediately obvious the enormous data requirements involved in the data acquisition, but a simple order of magnitude calculation provides an intuition for how much memory is required. In an earlier blog post by the chief scientist, Jun Korenaga, we learnt that the Shatsky is about the size of California. We will be making several trips across the Shatsky rise, and in total we will collect 3, 500 km of reflection data, which enable us to "see" into the crust beneath the ocean floor.

So what does 3, 500 km of MCS data translate to? Quick math: at 20 shots per km (because we shoot at every ~50 m), 3, 500 km of seismic survey will require 70, 000 shots. Each of these shots is recorded by the 468 channels, so we have 468 x 70, 000 = 32,760,000 shot traces. Now, as I mentioned, each channel samples data every 2 milliseconds and for each shot we have sampling continuing for 16 seconds (so 8,000 samples per trace). Four byte per sample will then gives the total memory required as: ( 32,760,000 * 8,000 * 4) bytes = 1,048,320,000,000 bytes. Large? Maybe not so much. This is actually 976 Gb (close to 1 tera bytes). I have a 6 Gb iPod touch, and if I were to store all the data on this model, I would require ~ 163 iPods. Yes, that's a lot of iPods. An interesting story is the history of the growth of memory capability for seismic acquisition, but that's for another day. Right now, I need to get back to my watch.

No comments:

Post a Comment