I/O

One of the most complicated aspects in dealing with 3-D datasets in general, and SEPlib90 datasets in particular, is I/O. With a simple SEPlib77 dataset you could read in n, o, d, and label from the history file and have a complete description of the dataset. A simple call to sreed would then bring in all of the necessary data. With SEPlib90 datasets, its 6 files, numerous permutations, and its ability to deal with irregularity the task is much more daunting. What sep_reed provides is a way back to the ``good old days''. A simple call

         sep_reed("in",sepdata)

will read in a SEPlib77 or SEPlib90, header or no headers, regular or irregular, traces in order or out of order dataset into the sep_90 type.

In addition with optional arguments the user can:

• read in a a subset of the dataset,
• to a subspace of a sep_90 type,
• read in only the headers,
• read only the `space', or grid dimensionality,
• pad with zeroes in traces that correspond to SEPlib90 grid cell holes.

The entire list of options for sep_reed can be found on the web.

Just as sep_reed simplified reading in datasets, sep_rite and finish_tags making writing datasets easier.

sep_rite provides in its simplistic form a two argument call:

         call sep_rite(tag,sepdata)

to write out any variety sep_90 dataset. In addition it provides numerous optional arguments to write out subsets or too subsets of a SEPlib dataset on disk.

3-D datasets are usually irregular and are not possible to keep in core. As a result we generally process the dataset in sections, reading only what we need and writing what we are finished with. When writing these datasets to disk, the total number of traces and headers is not known until the very end. finish_tags provides the necessary mechanism to deal with this problem. Every time a call to sep_rite is made the number of traces written to a given tag is recorded. A call to finish_tags writes out to all the output tags the total number of traces written.