We currently have a huge treasure trove of telescopic observations lying dormant on hard drives that few people can access. What if we could combine those observations into an easy-to-access database that not only provides unprecedented fidelity, but also access to changes in the night sky over time?
This is precisely the goal of the DeepSky Database, an effort to catalogue, distill, and retrieve 90 terabytes of observations from the Mt. Palomar observatory in San Diego, California over the past few decades. My work at Lawrence Berkeley National Laboratory with the Computational Cosmology group created the automated MPI/Linux pipeline that makes it possible.
The pipeline takes in user-input coordinates, retrieves the relevant observations, identifies classes of astronomical objects using machine learning techniques, compares the results to an internal astronomical catalogue, and aligns, scales, and skews the observations so they can be added together. When tens to hundreds of images are properly added, it is like multiplying the observation time by an order or two of magnitude, returning extremely detailed pictures at your fingertips.
You can access the database at http://deepskyproject.org, which runs off of supercomputers at NERSC, and you can learn more about it here and here. Recent work has used our database to present an unprecedented recording of a supernova evolving over time! You can watch the video here.