What's in an EM database?

During the development and the early operation of its TEMP-VEL technology, PetroMarker recognized the need for a more “clever” database than the ones in use so far.

Since 2007, PetroMarker has therefore chosen to work with the HDF5 data format.

The reason for picking this technology is its superiority in efficiently allocating both complex objects and high-density time series. 

However, PetroMarker does not use it only for data exchange purposes, but to support the workflow from acquisition to interpretation.  As such it is a key technology in creating quality controlled and processed data in a matter of minutes. 

By pushing the data format issue towards the data sources themselves, PetroMarker is able to remove bottlenecks and dependencies usually found in these systems.

PetroMarker has organized the data in two main directions: One is "horizontal" containing long time series on a per-unit basis, the other is "vertical" containing a system-wide view of the complete system between two events on a common timeline.  This redundancy is necessary because different processing approaches require different data sources.  To keep everything in one big file is not efficient and to split it in many parts with complex relationships is not maintainable either.  The case of the actual data format is secondary as we have tools to create and maintain HDF5 files as we move along.

PetroMarker's HDF5 files provide a natural naming scheme. This means that a receiver node is named "receiver" and a transmitter node is named "transmitter".  If there are multiples, the usual HDF5 naming appends an underscore with an index number, e.g. "receiver_2".  Any inter-relationship between the files is laid out in a reference-scheme similar to hyperlinks on the web.  These link the various files together within a project.  Of course, with a standard HDF5 viewer it looks like your ordinary HDF5 node, but with little additional machinery it provides full control of the data mapping between the various sources.  We have also built in features to maintain full history of the file.

PetroMarker never hardcodes HDF5 tree structures.  The HDF5 format and structure is created and maintained in other tools.  This saves a lot of development time and supports a dynamic and evolving process.

A typical PetroMarker survey contains about a hundred HDF5 files ordered in a simple, logical manner.  It is also possible to extend this to deliver both model files and processed data, or view them online with PetroMarker's unique web application d4.

LEGEND

1. Various sources of data during acquisition
2. Data are immediately aligned and formatted in a structured way
3. The resulting HDF5 files are supporting the rest of the workflow both offshore and onshore:

• All information are available on the www.
• QC is run on HDF5 files and results are easily verifiable.
• The HDF5 files are the basis for processing and inversion.
• The HDF5 files are delivered on a suitable media to the client.
• Other commercial and non-commercial tools have HDF5 readers, etc..Here.