While ground-based imaging is powerful due to its wide field of view, it is limited in a critical aspect. The earth’s atmosphere inhibits us because it both blurs out sources and reduces the amount of light reaching us on the ground. We see sources with with less detail from the ground, and can’t observe the faintest sources at all.
One solution to both these issues is to launch a telescope into space, which is exactly what has been done with the Hubble Space Telescope. Using Hubble, we can see globular clusters with enough detail to distinguish them unambiguously from contaminants, a task which can only be done in a probabilistic sense from the ground. Unfortunately, time on Hubble is very competitive, and even the largest cameras on Hubble have a drastically reduced field of view compared to what we see from the ground. One can compensate for the small field of view by tiling many images of distant galaxies, but given how tough it is to get time on Hubble, only a few such datasets exist.
I analyzed one such dataset in my first grad school paper. I examined a deep, six-pointing mosaic of the nearby galaxy NGC 3115. I performed both standard photometry of sources in the image as well as size measurements of resolved GC sources, requiring one to convolve modelled light profiles with the intrinsic point spread function of the instrument for several hundred sources.
Relevant Papers: Jennings et al. 2014a