The Challenge of Data Reduction for Multiple Instruments on the Stratospheric Observatory for Infrared Astronomy (SOFIA)

SOFIA, the Stratospheric Observatory For Infrared Astronomy, presents a number of interesting challenges for the development of a data reduction environ- ment which, at its initial phase, will have to incorporate pi pelines from seven different instruments developed by organizations around the world. Therefore, the SOFIA data reduction software must run code which has been developed in a variety of dissimilar environments, e.g., IDL, Python, Java, C++. Moreover, we anticipate this diversity will only increase in future generations of instrumentation. We investigated three distinctly different situations for performing pipelined data reduction i n SOFIA: (1) automated data reduction after data archival at the end of a mission, (2 ) re-pipelining of science data with updated calibrations or optimum parameters, and (3) the interactive user- driven local execution and analysis of data reduction by an investigator. These different modes would traditionally result in very different software implementations of algo- rithms used by each instrument team, in effect tripling the amount of data reduction software that would need to be maintained by SOFIA. We present here a unique approach for enfolding all the instrument-specific data reduction software in the observatory framework and verifie s the needs for all three re- duction scenarios as well as the standard visualization too ls. The SOFIA data reduction structure would host the different algorithms and techniques that the instrument teams develop in their own programming language and operating system. Ideally, duplica- tion of software is minimized across the system because instrument teams can draw on software solutions and techniques previously delivered to SOFIA by other instruments. With this approach, we minimize the effort for analyzing and developing new software reduction pipelines for future generation instruments. We also explore the potential benefits of this approach in the portability of the software t o an ever-broadening sci- ence audience, as well as its ability to ease the use of distri buted processing for data reduction pipelines.