Jeroen De Jong (Max Planck Institute for Extra-Terrestrial Physics (MPE)), Jeroen de Jong (MPE), Ekkehard Wieprecht (MPE), Juergen Schreiber (MPIA), Rik Huygen (KUL), Michael Wetzstein (MPE), Pierre Royer (KUL), Bart Vandenbussche (KUL), Katrina Exter (KUL), Roland Vavrek (ESAC), Beatriz Gonzalez (ESAC), Javier Diaz (ESAC), Jorgo Bakker (ESAC), Eckhard Sturm (MPE)
Abstract
We present the extensions to the Herschel Common Software System
(HCSS) which we wrote for simplifying the tasks in the pipelines of
the PACS instrument of ESA's Herschel Space Observatory. HCSS is
developed using Java technology by the Herschel Science Center and the
three instrument teams. In close collaboration with PACS users we made
many improvements for facilitating interactively processing the data,
while sharing the same Jython scripts with the Automatic Processing
which runs in an operational bulk processing environment. In this way
we can easily keep the pipeline up-to-date with user contributed
improvements.
Our most important extensions concern how to divide the products into
slices, and how to handle these slices in a transparent way for
users. Slicing is needed to divide the observation in scientific
blocks which require different processing, and to save memory since
data products can be several gigabytes. We show on this poster how we
extend the HCSS task and product classes to hide the handling of
slices for users and provide convenient handling of the data
products. This includes our classes for organizing and splitting up
the data, handling internal loops in tasks, memory management, meta
keyword handling and special convenience products.
Users who want to add a task to a PACS pipeline only have to extend
our augmented task classes and put their algorithm and parameters in
predefined methods. All this could in principle be applied to other
pipelines as well. Even C or FORTRAN code could be used via the Java
Native Access (JNA) interface.
Paper ID: P031
Poster Instructions
|
