Example data
<?xml version="1.0"?>
<name>line 101</name>
<xyzs_field_names>xcoord,ycoord,elevation,common depth point</xyzs_field_names>
123456.712346 123556.712346 123656.712346 123756.712346 123856.712346 ... [ 1000 floating point numbers ] </xcrd>
1234567.812346 1234667.812346 1234767.812346 1234867.812346 ... [ 1000 floating point numbers ] </ycrd>


Energy industry


This is a synthetic seismic dataset. It simulates a single 2D line after processing. (Field data have very different data structure because of the highly redundant nature of the acquisition). A 3D dataset might be composed of several hundred such lines, but for most purposes it is very likely that a single line is the unit of data that a Web service would be called on to deliver. That is, if one wanted to pass a full 3D dataset one would probably do it by repeated transmission of data like this. There is a small amount of noise added to these data, which otherwise represent a very simple earth model. The data start out with header information which specifies things like trace number, what location the trace corresponds to, and so on. It then has 1000*1501 numbers which are the traces themselves.

Here are a few of the many differences between this synthetic dataset and real seismic data.

  1. The large signals on these traces are much more coherent from trace to trace than real data.
  2. Real data have a lot more events that are not flat and are otherwise very complex.
  3. Noise in real data is not really white.
  4. Real data would have more information in the headers. If you're interested, the industry standard for headers is documented at http://w3.uwyo.edu/~seismic/sia/doc/segy_header.html.
  5. The numbers in the location data are much more regular and repetitive than would be the case in real data.