Cosmoglobe DR2. III. Improved modelling of zodiacal light with COBE-DIRBE through global Bayesian analysis

TL;DR

This work advances zodiacal light modelling by applying a global Bayesian Cosmoglobe DR2 framework to DIRBE data, leveraging external surveys (Planck HFI, WISE, Gaia) to jointly constrain ZL with astrophysical components. The authors implement a parametric IPD model (smooth cloud, three dust bands, circumsolar ring, and Earth-trailing feature) and convert densities into observable intensities via scattering and thermal emission, including a solar-centric excess component discussed elsewhere. Through a Gibbs-sampling–based joint fitting, they obtain updated ZL parameter estimates that yield cleaner ZL-subtracted maps and reveal significant differences from the classic Kelsall 1998 model, particularly in mid-infrared channels. The study highlights remaining degeneracies and residuals, and argues that future work combining higher-resolution data (AKARI, IRAS, SPHEREx) and improved sampling algorithms will be essential to reach an optimal Bayesian ZL model for infrared cosmology.

Abstract

We present an improved zodiacal light (ZL) model for COBE-DIRBE derived through global Bayesian analysis within the Cosmoglobe Data Release 2 framework. The parametric form of the ZL model is inspired by the original DIRBE model by Kelsall et al. (K98), but the specific best-fit parameter values are re-derived using the combination of DIRBE Calibrated Individual Observations, Planck HFI sky maps, and WISE and Gaia compact object catalogs. Furthermore, the ZL parameters are fitted jointly with astrophysical parameters, such as thermal dust and starlight emission, and the new model takes into account excess radiation that appears stationary in solar-centric coordinates as reported in a companion paper. The relative differences between the predicted signals from K98 and our new model are $\lesssim 3\%$ in the 12 and 25 $μ$m channels over the full sky. The zero-levels of the cleaned DR2 maps are lower than those of the K98 ZL Subtracted Mission Average maps by $\sim 30$ kJy/sr at 1.25--3.5 $μ$m, which is larger than the entire predicted contribution from high-redshift galaxies to the Cosmic Infrared Background at the same wavelengths. At high Galactic latitudes, the total RMS of each DR2 map is lower than the corresponding DIRBE ZSMA map of $\sim$ 80 \% at wavelengths 4.9--25 $μ\mathrm{m}$. Still, obvious ZL residuals can be seen in several of the DR2 maps, and further work is required to mitigate these. Joint analysis with high-resolution full-sky surveys such as AKARI, IRAS, Planck HFI, and SPHEREx will be essential both to break key degeneracies in the current model and to determine whether the reported solar-centric excess radiation has a ZL or instrumental origin. Thus, while the results presented in this paper do redefine the state-of-the-art for DIRBE modelling, it also only represents the first among many steps toward a future optimal Bayesian ZL model. (abridged)

Figures (15)

• Figure 1: Geometry of the first asteroidal dust band. (First row:) Slice through the $x$--$z$ plane of the number density, $n_{0}$, in heliocentric coordinates. The positions of the Sun and Earth are marked by orange and green dots, respectively. (Second row:) Observed instantaneous intensity plotted in Ecliptic coordinates, obtained by integrating the above figure along each line-of-sight. (Third row:) Same as above, but plotted in Galactic coordinates and mission averaged (MA) over nearly a full year of observations corresponding to the DIRBE scanning strategy. (Fourth row:) Difference between observed intensities as defined in the third row after changing the value of the ascending node, $\Omega$, by 5%. Similar plots for all components and parameters are provided in Appendices \ref{['sec:zodi-comps']} and \ref{['sec:param-atlas']}.
• Figure 2: Illustration of the basic sky maps involved in the ZL fitting algorithms adopted by the Cosmoglobe (left column) and K98 (right column) pipelines for one week of $25\mu\mathrm{m}$ observations, both adopting the K98 model. The basic data element in Cosmoglobe is the full sky signal, $I_{\nu}$ (top left), which is fitted with the full ZL model (K98 in this case), $Z_{\nu}$ (middle left), both modeled in time-domain. The $\chi^2$ used in the Cosmoglobe analysis minimizes the total signal-minus-model residual, $I_{\nu}-Z_{\nu}$ (bottom left). In contrast, the K98 pipeline used exclusively differences between weekly and full-season maps, both for the observed signal, $\Delta I_{\nu} \equiv I_{\nu}-\left<I_{\nu}\right>$ (top right), and the ZL model, $\Delta Z_{\nu} = Z_{\nu}-\left<Z_{\nu}\right>$ (middle right), where brackets indicate full-survey averages. Correspondingly, the final $\chi^2$ is defined through $\Delta I_{\nu} - \Delta Z_{\nu}$ (bottom right), and is by construction only sensitive to time-variable signals. The main advantage of the K98 approach is insensitivity to stationary sky signals, in particular thermal dust and CIB, while the main advantage of the Cosmoglobe approach is a much higher effective signal-to-noise ratio, both to ZL parameters and zero-levels, as seen by comparing the two bottom panels.
• Figure 3: Reduced $\chi^2$ as a function of Powell likelihood evaluation count for one single pre-production Gibbs chain, showing the burn-in phase. Each discrete jump indicates the start of a new Gibbs sample, which is initialized on a new random point that is close to the previous iteration. The following systematic decline within each main Gibbs iteration indicates the non-linear optimization performed by the Powell algorithm. The solid dark region corresponds to a large number of highly sub-optimal parameter trials.
• Figure 4: Reduced $\chi^2$ as a function of the temperature at 1AU, $T_0$, for the same run shown in Fig. \ref{['fig:powell_chisq_iter']}. Each curve shows the full set of parameter trials within one single main Gibbs iteration (or Powell call), and different colors indicate different Gibbs iterations. Redder colors are earlier in the chain.
• Figure 5: Subsample of the pre-processed TOD used in this analysis for all ten DIRBE bands. The time-streams show approximately one rotation of the COBE satellite which includes two crossing of the Galactic and Ecliptic planes.