| Name: | Sougata Sarkar |
| Affiliation: | Indian Institute of Science |
| Conference ID : | ASI2024_532 |
| Title : | WNM and CNM identification in GARCIA galaxies using the Gaussian Mixture Model (GMM), an unsupervised machine learning method |
| Authors : | Sougata Sarkar, Maitraiyee Tiwari, Prerana Biswas, Nirupam Roy |
| Authors Affiliation: | Sougata Sarkar (Indian Institute of Science, Bangalore - 560012, India)
Maitraiyee Tiwari (University of Maryland, USA)
Prerana Biswas (Indian Institute of Science, Bangalore - 560012, India)
Nirupam Roy (Indian Institute of Science, Bangalore - 560012, India)
|
| Mode of Presentation: | Poster |
| Abstract Category : | Galaxies and Cosmology |
| Abstract : | Multiple thermal phases of gas are known to exist in Interstellar Medium. HI gas can manifest in distinct thermal states due to various heating and cooling processes in the ISM caused by the feedback from massive stars during their lifetime and after death (supernovae). Molecular clouds comprising molecular hydrogen (H2) represent the coldest phase of the ISM, with temperatures ranging from 10-20 K. The Cold Neutral Medium (CNM) comprising atomic hydrogen (HI) has a temperature range of 20-100 K. In contrast, the Warm Neutral Medium (WNM) has much higher temperatures ranging from 5000-8000 K and much lower densities when compared to the CNM. Given their distinct temperature ranges, both CNM and WNM leave their unique imprints on the hydrogen line spectra. Narrow spectral widths are associated with CNM, while wide spectral widths are associated with WNM, making their identification possible. Patra et al. 2016b and Biswas et al. 2020 used the Gaussian decomposition method to differentiate between WNM and CNM by decomposing the H I line-of-sight velocity profiles into multiple Gaussians and correlating them to kinetic temperatures. Several limitations are associated with using this technique: reduced sensitivity with SNR, inability to perform region-wise averaging without prior knowledge of cold and warm regions and high uncertainties associated with the identified components. To this end, we have used the Gaussian Mixture Model (GMM), which is an automated technique to identify distinct physical structures in the ISM to distinguish between the warm and cold ISM phases of GARCIA (GMRT archive atomic gas survey) galaxies. GMM, when applied to a velocity-resolved dataset, finds clusters of spectra that have similar profiles to each other. Through this talk, I will present the first results of using this technique to identify the CNM and WNM in GARCIA galaxies. |
