LSST Cadence Optimization White Paper in Support of Observations of Unresolved Tidal Stellar Streams in Galaxies beyond the Local Group

Please contact Seppo Laine, Caltech/IPAC, seppo@ipac.caltech.edu, with any questions about this white paper.

This white paper addresses:

1. 1.

   Science Category: Milky Way Structure and Formation: Exploring the Faint Surface Brightness Universe.

2. 2.

   Survey Type Category: Wide–Fast–Deep Survey.

3. 3.

   Observing Strategy Category: Targeting areas of sky off the Galactic plane.

Previous deep, wide-area photometric surveys have revealed a large number of faint stellar substructures (“streams”) around galaxies like the Milky Way (MW), resulting from the tidal disruption of lower-mass galaxies in “minor mergers” or previous major mergers. While detailed studies of resolved streams around the MW and M31 imply a dynamic hierarchical accretion history, consistent with $\Lambda$CDM cosmological galaxy formation models (e.g., Bullock & Johnston, 2005; De Lucia & Helmi, 2008; Cooper et al., 2010, 2013; Pillepich, Madau, & Mayer, 2015; Rodríguez–Gómez et al., 2016), we need to study a much larger sample of galaxies to test whether the merging histories of MW and M31 are typical of galaxies in their mass range (e.g., Mutch, Croton, & Poole, 2011; Morales et al., 2018).

A crucial ingredient in testing whether the merging histories of the MW and M31 are typical (consistent with $\Lambda$CDM cosmological galaxy formation models) is the acquisition of adequately deep images, as the majority of the predicted tidal stellar streams have surface brightnesses in the R-band fainter than about 29 AB mag arcsec^-2. While a few deep imaging surveys of the outskirts of local galaxies have recently been completed (e.g., Tal et al., 2009; Martínez Delgado et al., 2010; Ludwig et al., 2012; Duc et al., 2015) or are ongoing (Abraham & van Dokkum, 2014), the majority of nearby galaxies have not been observed down to surface brightnesses needed to detect streams from ancient minor mergers.

By focusing on nearby spiral galaxies with diffuse-light overdensities, more than 50 previously unknown stellar structures in galaxies at distances $<$ 80 Mpc have been discovered so far (Martínez-Delgado 2018). The morphologies of the diffuse-light structures include “great circle”-like (such as are seen around the Milky Way) streams that roughly trace the orbit of the merging satellite galaxy (Sanders & Binney, 2013), isolated shells, giant debris clouds, jet-like features, and large diffuse structures that may be old, phase mixed remnants of merged companions (see Figure 2 for examples). Again, very similar features are seen in cosmological simulations of minor mergers (Johnston et al., 2008; Cooper et al., 2010). Although it appears in many cases that the progenitor companion has been completely disrupted, a few examples (Martínez Delgado et al., 2012, 2015; Amorisco, 2015) show surviving cores of merging companion galaxies, often exhibiting long tails departing from the progenitor satellite.

One of the main objectives of a conceivable LSST stream survey among MW-like galaxies is the comparison of the observed frequency and the parameters of the streams, such as their spatial coherence, length, width, inclination, morphology, color and surface brightness to those seen in cosmological simulations. These models suggest that remnants from mergers 0–8 Gyrs ago are still visible as substructures in the halos of nearby galaxies. Information about the streams comes from 1) morphology of the low surface brightness emission, as Johnston et al. (2008, their Fig. 3) and Hendel & Johnston (2015) have shown that different morphologies of tidal debris occupy different regions in the accretion time vs. orbital eccentricity/energy plane; 2) the progenitor’s luminosity; 3) stellar population of the stream (to infer its stellar mass); 4) shape and width of the streams (to obtain the dynamical properties of the progenitors; Johnston et al., 2001; Erkal, Sanders & Belokurov, 2016); 5) color of the stream (to determine whether a minor or major merger was involved); 6) mass and morphological type of the host galaxy (to obtain the frequency of mergers); and 7) (combined with the previously listed information) surface brightness that can be used to time the epoch of the merger (and thus the rate at which new streams are being formed in the local Universe; Johnston et al., 2001).

The halos in the simulations of Bullock & Johnston (2005) typically have about two streams brighter than 30 AB mag arcsec^-2. However, the majority of substructures are at surface brightnesses fainter than 30 AB mag arcsec^-2. Our current inability to see the fainter streams (corresponding to either earlier merger epochs or lower mass progenitors) implies that currently the merger history that we study in galaxies beyond the Local Group is strongly biased towards the most recent (the last few tens of percent of mass accretion) and/or most massive minor merger events. Thus we are only sensitive to the most metal-rich populations, as validated by studies of resolved stars around M31 (McConnachie et al., 2018) and Cen A (Crnojević et al., 2016). Close to the center of a galaxy (R_proj $<$ 30 kpc) the substructure is most likely generated by the most massive merged satellite galaxies, as dynamical friction will have brought them quickly to the central regions. Therefore, our current view of tidal streams in nearby galaxies is highly biased towards the most massive minor mergers which are relatively rare for MW-like galaxies.

Studying the frequency, mass ratios and stellar populations of minor mergers with the help of stellar tidal streams in external galaxies will also provide critical input on a number of open astrophysical questions, such as 1) heating and thickening of the host galaxy disk by satellite mergers, including the frequency and impact of low orbital inclination satellites, most recently showcased by the Gaia-Enceladus galaxy remains in the MW (Helmi et al., 2018); 2) as discussed above, the hierarchical build-up of the primaries; 3) dark matter halo shapes, when combined with N-body simulations and 4) the fundamental nature of dark matter. Regarding this last point, there are alternative candidate particles to cold dark matter that are well motivated from particle physics and that result in radically different properties for astronomical objects on the small scales probed by stellar streams. One of the currently popular alternatives to CDM is sterile neutrinos with a mass of a few keV, which behave as warm dark matter (WDM). On the large scales probed by the cosmic microwave background and large-scale structure, they are indistinguishable from CDM. But on small scales they are very different. In particular, they require a cutoff in the mass function of dark matter halos on scales of about 10⁹ M_⊙ for particle masses of interest. It is likely that WDM models predict different properties for tidal stellar streams, although this is still to be verified with high-resolution simulations. Other particle models, such as certain types of self-interacting particles, also predict a cutoff in the halo mass function, which may also lead to differences in the properties of the streams.

The spectral energy distributions (SEDs) of the merged galaxies can be best studied by combining the LSST observations with data from near-IR wavelengths. WFIRST will provide a survey of several deep fields efficiently and will provide an ideal complementary data set near the peak of the stellar SED. The optical images from Euclid may also be useful. The combined LSST/Euclid/WFIRST data set will provide a broad wavelength baseline for the estimation of the ages, metallicities and masses of the stellar populations of disrupted companions, producing valuable constraints on the minor merging history in CDM models of hierarchical galaxy formation.

The most obvious heuristic is the total depth achieved during the survey which is directly related to the number of detections of low surface brightness streams in integrated light. Based on the deepest imaging observations ever taken, 8 hours on-source using the 10.4-meter Grant Telescopio de Canarias (GTC) telescope and with an average seeing of 0.8–0.9 arcseconds (Trujillo & Fliri, 2016), we have made an approximate estimate of the expected low surface brightness limit that the LSST can provide after the scheduled 825 visits to the same sky location. This corresponds to a total amount of time on source of 3.44h. The expected surface brightness limits will be (3$\sigma$; 10$\times$10 arcsec² boxes): 29.9 ($u$), 31.1 ($g$), 30.6 ($r$), 30.1 ($i$), 28.7 ($z$) AB mag arcsec^-2. Each visit that consists of 30 seconds on-source will correspond to the following limits (3$\sigma$): 26.6 ($u$), 27.8 ($g$), 27.3 ($r$), 26.8 ($i$), 25.4 ($z$) AB mag arcsec^-2. If the WFD survey depth is increased by 10%, our total detection limit goes up only by 0.05 mag arcsec^-2.

As suggested by the state-of-the art cosmological simulations (e.g., see Figure 1), these surface brightness limits would enable us to detect new stellar tidal streams in diffuse or integrated light around 80–90% of our sample of a few thousand galaxies out to several hundred Mpc. For obtaining stellar population diagnostics, we require at least a 5$\sigma$ detection in a minimum of three bands. We estimate that with the current WFD survey parameters we would be able to perform stellar population diagnostics in a few thousand stellar tidal stream systems in integrated light.

Obtaining quantitative information from measurements of low surface brightness emission requires reducing data in a completely different way from that of point source driven data. One requirement we impose on the data is that individual images should be made available, not just coadds, as scattered light is much easier to remove from individual frames than from image stacks. In addition to the removal of scattered light, there are several other considerations for low surface brightness measurements. While the LSST data reduction pipeline masks high surface brightness sources, the effects from these sources and bright stars can masquerade as low surface brightness emission because of the extended PSF. We propose to use software such as imfit (Erwin, 2015) to remove the stellar envelope from bright host galaxy emission by modeling the host galaxy surface brightness model assuming it is a simple Sersic profile.

We will also select images for the deepest image stacks based on sky brightness at the time of the observations, taking into account the vicinity (and phase) of the Moon to the observed area of the sky, as well as experiment with using all data, regardless of the Moon phase.

Another consideration for the detection of low surface brightness emission is stellar aureole (from ice particles in thin cirrus clouds in the atmosphere). We propose to develop techniques to remove this emission from the LSST images.

To estimate the limiting surface brightness in the image stacks, we intend to run simulations by injecting substructure into the simulated images to test recovery as a function of surface brightness.

Once the images have been reduced, and the erroneous sources of low surface brightness emission have been removed, we will use special techniques to detect tidal stellar streams in diffuse (or integrated) light, such as software adapted from HSC–SSP (Subaru Hyper Suprime-Cam Strategic Program; Kado–Fong et al., 2018). We will use adaptive smoothing techniques (similar to Zibetti, 2010), NoiseChisel and iterative unsharp masking techniques that separate structure in the image by spatial frequency. While we will inspect the images by eye to detect features, we will also develop algorithms for automatic detection and morphological classification of streams, which are expected to be available by the time LSST starts operations (c.f. Hendel et al., 2018).

Finally, we will perform an estimation of the limiting surface brightness (3$\sigma$) in the stacked images and will make the re-reduced stacked deep images available to the community as soon as the basic image reduction has been completed.