A geostatistical analysis of multiscale metallicity variations in galaxies [II]: Predicting the metallicities of Hii and diffuse ionised gas regions via universal kriging

TL;DR

This paper introduces universal kriging, a geostatistical method, to accurately reconstruct galaxy metallicity maps from limited data, improving predictions for diffuse ionised gas regions and outperforming traditional gradient interpolation.

Contribution

The paper applies universal kriging to galaxy metallicity mapping, demonstrating its effectiveness in predicting metallicities in DIG regions and outperforming gradient-based methods.

Findings

Kriging captures spatial correlations up to 1.2 kpc.

Predictions outperform gradient interpolation in cross-validation.

Systematic offsets in DIG metallicities are within ±0.1 dex.

Abstract

The metallicity of diffuse ionised gas (DIG) cannot be determined using strong emission line diagnostics, which are calibrated to calculate the metallicity of Hii regions. Because of this, resolved metallicity maps from integral field spectroscopy (IFS) data remain largely incomplete. In this paper (the second of a series), we introduce the geostatistical technique of universal kriging, which allows the complete 2D metallicity distribution of a galaxy to be reconstructed from metallicities measured at Hii regions, accounting for spatial correlations between nearby data points. We apply this method to construct high-fidelity metallicity maps of the local spiral galaxy NGC 5236 using data from the TYPHOON/PrISM survey. We find significant correlation in the metallicity of Hii regions separated by up to 0.4-1.2 kpc. Predictions constructed using this method were tested using cross-validation in Hii regions, and we show that they outperform significantly interpolation based on metallicity gradients. Furthermore, we apply kriging to predict the metallicities in regions dominated by DIG emission, considering seven additional spiral galaxies with high resolution (<100pc) metallicity maps. We compare kriging maps to DIG metallicities computed with novel ionisation corrections, and find that such corrections introduce a systematic offset of up to $\pm0.1$ dex for any individual galaxy, with a scatter of 0.02-0.07 dex for the sample. Overall we recommend universal kriging, together with a calibrated geostatistical model, as the superior method for inferring the metallicities of DIG-dominated regions in local spiral galaxies, demonstrating further the potential of applying geostatistical methods to spatially resolved galaxy observations.