The PS1 telescope has been running a wide-field NASA funded survey for near earth objects since early 2014, with PS2 coming online later in the year. The first ATLAS telescope has been on the sky since late 2015, and the second telescope will be active in late 2016. QUB are using these as test streams to develop LSST processing software, specifically Level 3 data products.

The LSST transient server work is now starting to test LSST Qserv database software (hosted in Edinburgh) and we are testing algorithms on this live, daily data stream.

The difference image detection catalogues are ingested into an SQL database, aggregated into objects and filtered using machine learning. Filtered objects are presented to human scanners in the form of a web interface. The difference detection database sizes are currently ~several TB, with 10s of millions of detections ingested each night. We envisage similar processes for LSST.

In addition to finding transient objects, the ATLAS telescope is generating 1000 lightcurve points per year for a billion stars - so potentially 1 trillion detections per year need to be inserted into a database at a rate of at least 64,000 per second (assuming 12 hours ingestion time daily). Experience of processing and storage of the ATLAS data is a good precursor for future LSST variable object work.

As we gain practical experience in the data management flow, the UK community can assist by developing science requirements for LSST.