9.2 PipeLine

Chapter 2 discussed the need for an automated method of reducing observational data and showed that the desire for such a system already existed. The theory of data reduction was then discussed. Chapter 3 revealed that whilst the need for automated data reduction existed very few such systems did exist, however programs existed to perform individual sections of data reduction. The robotic LT was also introduced as a prime candidate in need of an automated reduction system.

In Chapter 4 a prototype PL was designed and implemented. Through this exploratory work the fundamentals of automating data reduction were learned. The prototype was used to analyse low mass star data producing the first high quality main sequence for an open cluster all the way down to the hydrogen burning limit. In Chapter 5 the lessons learnt from the prototype were implemented in a production PL designed for the LT. This new PL was able to overcome several of the problems faced by the prototype (e.g., inputting a WCS, automatic airmass curve creation). The results produced by the production PL were verified against the prototype, this showed that the production PL was producing quality data.

The field of astronomical data analysis software and systems has experienced phenomenal growth in the last few years, new methods, better techniques and faster computers are continually being developed. As such no reduction software will ever remain at the forefront of the field unless it evolves to embody new ways. The LT PL was specifically designed to allow for its continuing growth, its infrastructure is based in a modular design to allow pipes to be removed and upgraded with the minimum of interference to other sections of the PL. This evolving nature is the PLs true legacy: whilst the methods it uses have been tried and tested they may be superceded by new techniques, or different implementations of the existing algorithms, the PL infrastructure will always remain at its core.

Even now there are possible upgrades to the PL; The SExtractor software is currently being re-written into a software library, removing the current redundant routines will speed PL throughput. In order to maintain the modular nature of the PL code some duplicate code was written into the individual pipes. Further subdividing each module can be achieved by converting this repeated code into a callable library, enhancing the upgrading options of the PL. The Solaris operating system employed by the PL allows “dual coding”, that is one piece of written code may contain both FORTRAN and C. Since the photometry pipe is written in FORTRAN it could be advantageous to employ this system.

The LT employs rudimentary weather monitoring equipment, it is planned to update this to provide a much more detailed and overall weather reporting system. By feeding this data into the PL it may be possible to enhance the indication of non-photometric frames, this could aid in the construction of airmass curves. The removal of rogue data points from plots is still problematic and one of the most difficult procedures to automate.