As part of a reverse engineering effort, I am trying to develop data flow models for various mainframe batch JCL procedures used by the mainframe application I support.  There are so many types of models to choose from I'm not sure which type of modeling technique would best depict the flow of a mainframe job procedure.  I need to depict the sequential flow of the datasets created, read, updated or deleted through each procedure step.   I couple thoughts I had was either using the structure chart model, or UML sequence diagram.

Has anyone modeled these types of data flows and can recommend a best practice modeling technique?