A similar interactive approach can be used for many of the pre- and post-imaging
steps now done at SEP using batch processes. Instead of reading in and processing
an entire volume, only the portions of the dataset needed for the
current view need reside in memory. This is a subtle but important difference.
Processes such as NMO operate on a single trace, but doing a velocity scan requires
all of the traces at a given CMP location. As a result each routine must have an
awareness of what range of data is required to display a given subsection. As the number of
processes increases the
application must be able to keep track of a potentially expanding tree
of data required to read into main memory to produce a
given view.
![[pdf]](icons/pdf.png){kind=icon}