Statistical Models of the Spatial, Kinematic, and Chemical Complexity of Dust

Abstract

Dust in the interstellar medium (ISM) impacts almost every astronomical observation. In this thesis, I describe how we can characterize the distribution and properties of dust based on its contribution to images and spectra using statistical techniques. In imaging, filaments of gas and dust can complicate measuring the flux of stars (photometry). I introduce a method to correct photometry on structured backgrounds that marginalizes over realistic estimates of the ISM and respects its non-Gaussian structures. Applying this method to the DECam Plane Survey (DECaPS2), I produced the largest (3.32 billion sources) photometric catalog, from which we have built the highest angular resolution (and one of the farthest reaching) 3D dust maps to date. Turning to spectroscopy, I show that careful component separation techniques are required to achieve precision measurements of diffuse interstellar bands (DIBs) in Gaia DR3 RVS and APOGEE DR17 spectra. The resultant APOGEE DIB catalog realizes the promise of DIBs as precision kinematic tracers of the ISM, that also have distance upper limits, and provides a tantalizing view of the chemical diversity in the NIR dust spectrum. In total, this thesis describes scalable algorithmic advances to the processing of astronomical images and spectra that will be essential to make full use of upcoming surveys such as those being conducted by Roman, LSST, and SDSS-V.