Tuning the Legacy Survey of Space and Time (LSST) Observing Strategy for Solar System Science

The SSP pipelines (Myers et al. 2013; Jurić et al. 2020) will link transient sources from the nightly visits into "tracklets" (potential linkages in the same night using linear extrapolation). SSP will identify new moving objects by attempting to link together three tracklets from within a 15-day window onto a heliocentric orbit. The current baseline LSST object discovery guidelines require pairs of observations on three separate nights, within a window of 15 days as the design goal and 30 days as the stretch goal; the 15-day requirement is a confident lower limit, but a 30-day window is a reasonable extension that is also useful to consider. Thus, the basic discovery metric searches for precisely this: pairs of observations on at least three different nights within 15 or 30 days, using the probabilistic detection value to determine what is visible or not. ⁴³ The metric allows for setting the minimum and maximum time separation between the individual visits in each pair; in the default configuration, the minimum time separation was set to 0 minutes, ⁴⁴ and the maximum time separation was set to 90 minutes, corresponding to the approximate limits suggested by early expectations for configuration for the solar system processing pipelines and very widely bracketing the typical expected separation of visits. The overwhelming majority of visits in the survey are acquired in pairs with a separation of 22–30 minutes (depending on the details of the survey configuration); the pairs of visits are usually acquired in "adjoining" filters (i.e., g and r or r and i visits for a pair); and, coupled with the large FOV of Rubin, most although not all observations of an object are followed up by a second observation in the same night. It is also helpful to consider objects that could be discovered via more traditional methods of identifying four observations on the same night (i.e., "quad detections"). This is particularly useful when considering observations of near- or interior-to-Earth asteroids within the special near-Sun twilight microsurvey, where observations are purposefully obtained in quads in order to secure identifications of these rapidly disappearing asteroids. If the observations of a given orbit−H combination meet the required criteria at least once, the object is considered "discovered"; to compare the results across different simulations of survey strategy, the discovery completeness is reported at both a bright and faint H value for each population. More details about the discovery metrics are presented in Jones et al. (2018), including more background about the potential for false-positive discoveries. In short, we do not expect a significant number of false-positive detections, regardless of survey strategy choices, with the criteria of three pairs of detections over a window of either 15 or 30 nights; this is due to a range of factors, including the low fraction of false-positive detections coming from difference imaging and the low likelihood of pairs of detections on three separate nights aligning within expected residuals for initial orbit determination.

The various survey strategy simulations and populations expose some basic trends:

• 1.  
  Discovery completeness for slow-moving populations, such as TNOs, depends strongly on the total area included in the survey. Because these objects move so slowly year over year and are relatively "easy" to discover via linking, the footprint itself is the most important consideration of the survey strategy, particularly for the brighter TNOs. The completeness for the fainter TNOs can also vary slightly depending on which filters are paired together in visits and whether the most sensitive filters are used often enough within the window.
• 2.  
  Discovery completeness for fast-moving populations, such as NEOs, depends more strongly on the number of visits per pointing. Since NEOs travel across much more of the sky on the timescale of the survey, the footprint is not as much of a constraint as for TNOs. However, the total number of visits in the survey is relatively constant with different survey strategies, and so the footprint influences the number of visits per pointing and thus the typical cadence of those visits. Fainter NEOs in particular may only be visible for a short period of time; thus, more visits per pointing result in a higher likelihood of an object having observations suitable for discovery, and so a higher discovery completeness. For the brightest NEOs, the footprint weighs in as well, as covering more sky results in discovering more NEOs.
• 3.  
  Intermediate populations, such as MBAs, fall in between these extremes. In general, we find a threshold number of visits per pointing results in good completeness for a given population, and this threshold increases as the H value being evaluated gets larger and/or the population includes more small semimajor axis or high-inclination or high-eccentricity orbits.
• 4.  
  The Jupiter Trojans show stronger variability with some kinds of survey strategy changes that include changes in the timing of observations. Some survey strategies focus visits on particular regions of the sky in particular years, such as in the rolling cadence. These variations can result in a higher or lower sampling of the Jupiter Trojan population depending on the timing of visits, as these asteroids are both more spatially constrained and moving across the sky.
• 5.  
  More relaxed discovery criteria result in more discoveries, but with similar trends. For example, 30-day windows perform about 2%–5% better than 15-day windows for fainter objects, depending on the population (brighter objects show little difference). However, these different criteria follow similar trends between survey strategies, meaning that evaluating 15-day windows shows similar preferences in survey strategy to evaluating 30-day windows.

Inner solar system objects have the potential to be subjects for sparse light-curve inversion, inferring the shape of the asteroid from photometric measurements over a wide range of viewing geometries as suggested in LSST Science Collaboration et al. (2009; e.g., Hanuš et al. 2011; Ďurech et al. 2016; Muinonen et al. 2020). The light-curve inversion metric evaluates the suitability of a set of observations for this process. The evaluation is based on the phase curve and ecliptic longitude coverage provided by the observations, as well as the overall number and S/N of each observation, considering observations in a single filter at a time. The ecliptic longitude range of the observations must be more than 90° ecliptic longitude and cover more than 5° of phase angle, as a proxy of the required range of viewing geometries. Further, there must be more than a threshold value of S/N-weighted observations, equivalent to about 50 S/N = 100 observations or 250 S/N = 20 observations, all in the same filter, in order to provide enough photometric measurements. Like all other metrics within MAF, the rotation of the asteroid and its impact on the photometric measurements is not considered; presumably this would be part of the light-curve inversion process. If all conditions are met, then light-curve inversion is at least potentially likely; thus, this metric provides a likely upper limit on the fraction of the population for which light-curve inversion may be possible. This is evaluated per orbit−H combination, and then the fraction of the population (at a bright and fainter H value) is reported. Outer solar system objects never achieve the required range of viewing geometries, and objects where the nucleus is shrouded with coma such as active comets are also not good candidates; this metric is not evaluated for these populations.

This metric is very sensitive to the number of observations per pointing, but also to the cadence of those observations. Generally, we find a trend across the simulations that the light-curve inversion results track in a similar sense for all of the inner solar system populations, with NEOs being least sensitive to survey strategy variations, followed by PHAs, then MBAs, and finally Trojan asteroids showing the most variation in metric results as survey strategies change.

There are several metrics relating to determining colors for the small body population members, tailored for inner solar system or outer solar system objects. As the LSST will not obtain instantaneous colors, each of these metrics also includes some requirement on measuring a light curve.

For the inner solar system, the color light-curve metric evaluates the number of S/N-weighted observations per bandpass to evaluate whether the color could be determined in that bandpass. Essentially, this could be translated to fitting the light curve in each bandpass alone and then combining these light curves to evaluate the color. The equivalent of 40 S/N = 5 detections or 10 S/N = 20 detections per filter are required, but the more extensive requirements that relate to achieving a range of viewing geometries for light-curve inversion are not, and no limitation is set on when the observations are acquired. The specific number of detections needed is based on an estimate of the amount of data that would be sufficient to measure basic light-curve, color, and phase-curve parameters with scientifically meaningful uncertainties. Although work is still needed to use the sparse LSST-like cadence to determine these parameters, a preliminary assessment suggests that 20–40 observations per color should be sufficient. While phase curves are also necessary for this analysis, we elected to keep the metric simple by not requiring a particular spread in phase angles. In practice, almost any cadence will produce sufficient constraint on the phase curve to allow for colors to be determined for the vast majority of objects. The summary values reported are the fraction of the population (at a given H value) for which two specific colors (g − r or g − i plus g − z or g − y), three specific colors (g − r, r − i, i − z or r − i, i − z, z − y), four colors (g − r, r − i, i − z, z − y), or all five colors (adding u − g to the four-color set) are potentially determined.

For the outer solar system, a slightly different color light-curve metric evaluates the number of observations reaching a minimum threshold (S/N ≈ 5). This metric requires at least 30 observations in a "primary" bandpass and then 20 observations in the additional bandpass(es). This is equivalent to assuming a light-curve fit in the primary bandpass with additional observations in the secondary bandpass serving to help fit the light curve and color, possibly simultaneously (such as would be possible with multiband Lomb–Scargle fitting). The summary values reported are the fraction of the population (at a given H value) for which one, two, or more colors can be fitted, without restrictions on which bandpasses are used.

As described above in Section 2.2, the most accessible and up-to-date orbital and absolute magnitude distributions have been used to model the expected LSST solar system detections. The physical and orbital properties of the modeled synthetic small bodies are driven by observational data, but the LSST cadence simulations do have to make some assumptions about these small body populations. This is particularly true on the smallest size scales that have not been very well probed by past wide-field surveys. The distribution of different surface types applied to the various simulated small body reservoirs will also impact the apparent magnitude of the synthetic objects in the various optical filters. Additionally, we have to make simplifying assumptions about active objects. We assume that all comets will generate dust coma with the same relation applied to calculate the observed apparent magnitude, and the effects of cometary outbursts are ignored. In addition, rotational brightness variations due to shape or uneven surface albedo are not accounted for in these simulations. Thus, the exact number of solar system minor planets found by LSST will differ from that "discovered" in the simulations explored in this paper because of these choices.

The smaller solar system minor planet populations, such as the main-belt comets (MBCs), Jupiter-family comets (JFCs), sungrazing comets, Neptune Trojans, and Centaurs, have not been simulated for this work. For the MBCs, sungrazing comets, and JFCs this is partly due to having to develop a representative activity model. We can instead use the populations that are simulated in the rubin_sim simulations as proxies to help inform what the impact of various cadences might be. Simulated survey strategies that will improve the metrics for MBAs and NEOs will also likely enhance the discovery and monitoring of MBCs and JFCs. Cadences that improve the chances of finding near-Sun 'Ayló'chaxnims will likely increase the LSST discovery rate of sungrazing comets, like the Kruetz family. As the Centaurs reside in the middle solar system, the impacts on the Centaurs can be extrapolated using the TNO and MBA simulation metrics.

No ISOs were simulated for this work. With only two ISOs known to date (Meech et al. 2017; Borisov 2019), the characteristics of this population are currently unconstrained. Long-period comets are distributed across the sky with a much larger range of ecliptic latitudes compared to the MBAs and TNOs, due to the effects of passing stars and Galactic tides that shape the Oort Cloud into a shell rather than a flared disk shape (Everhart 1967; Fernández 1997; Francis 2005; Higuchi et al. 2007; Brasser et al. 2010; Dones et al. 2015; Vokrouhlický et al. 2019; Higuchi 2020, and references therein). Recent predictions by Engelhardt et al. (2017), Seligman & Laughlin (2018), and Hoover et al. (2022) suggest that LSST ISO discoveries will cover a wide range of ecliptic latitudes and heliocentric distances similar to long-period comets. Thus, we assume that the trends seen for the simulated LSST OCC discoveries can provide some broad guidance for how the cadence decisions will impact LSST ISO discoveries. Like 1I/'Oumuamua, which was discovered at 0.22 au (Meech et al. 2017) moving at 62 per day, a subset of ISOs discovered close to Earth will on short timescales (≲10 days) look similar to NEOs (e.g., Cook et al. 2016). Therefore, the NEO metrics are also insightful for gauging the potential impacts to the ISO discovery rate.

The solar system MAF metrics assume equal detection efficiency across all areas of the survey footprint (even near the plane of the Galaxy, where stellar crowding may be a factor). Rubin Observatory's data pipelines will detect solar system bodies using difference imaging. Templates representing the static sky will be subtracted from the nightly images, and what remains will be a variable, transient, or moving source. This will help significantly in detecting solar system objects in regions of high stellar density, but stellar crowding will likely cause some decrease in the efficiency of Rubin Observatory's difference image analysis (DIA) and SSP pipelines. The MAF solar system metrics are likely overly optimistic near and in the Galactic plane, where stellar crowding is the highest. This should be kept in mind when examining the cadence simulations modifying the LSST Galactic plane observing strategy.

The Rubin scheduler aims to take image pairs, each night per pointing, to facilitate the identification of moving solar system objects (Ivezić & the LSST Science Collaboration 2013). The time between these repeat observations is a tunable survey parameter. The Rubin SSP pipelines (Myers et al. 2013; Jurić et al. 2020) require motion within a single night for initial discovery. Transient sources that appear stationary between the two images taken on the same night will not be included in the daily tracklets that the SSP algorithm will try to link with tracklets from previous nights. The Rubin SSP pipelines as currently planned will not be able to detect bodies beyond ∼100–150 au (see Section 4.4.1 for a detailed estimate), but other search algorithms will likely be developed by the wider community to search for very slow moving objects in the LSST data. The MAF solar system discovery metric does not account for SSP's slow motion limit. The only metrics really impacted by this are the estimated TNO discoveries. As long as the median separation between the observations is similar for a set of cadence simulations, then the output from the discovery metrics can be compared. We note that some care must be taken when considering the impact of varying the time separation between repeat observations, and we refer the reader to Section 4.4.1 for further discussion.

There are currently no MAF metrics that measure how precise small body orbital predictions and characterization will be based on Rubin Observatory observations. The accuracy of the orbits of moving objects is primarily driven by the observational arc length. There were no reasons to consider the observational arc length separately with a dedicated MAF metric because all of the observing strategy options currently being considered as part of the LSST cadence optimization exercise (see Section 3) have repeat coverage of the entire LSST footprint over several years. This should be sufficient for the needs of the majority of astrometric and dynamical solar system science cases. We also note that if a cadence option not covering the entire sky over the majority of the 10 yr time span were evaluated, it would be undesirable for other science cases such as proper-motion measurements.

The likelihood of having satellite streaks and glints present in LSST images is increasing with every satellite constellation launch (e.g., Starlink, Project Kuiper, and OneWeb). The impact of future satellite constellations is not currently taken into account by the metrics. We discuss the potential impacts of the ongoing industrialization of the near-Earth environment in Section 5.4.

Keeping these caveats in mind, the LSST cadence simulations and the MAF metrics can be used to explore the impact of various changes to the LSST observing strategy. Some care is required in examining certain families of simulations. Overall, by adopting the same synthetic small body populations for each of the cadence simulations and focusing on the relative change in the MAF metrics compared to the baseline survey, we can still gain a good understanding of the impact caused by tuning various LSST observing parameters.

The WFD by its design requirements is meant to cover the majority of the sky in the southern celestial hemisphere below 0° decl. (Ivezić & the LSST Science Collaboration 2013), but the Simonyi Survey Telescope is capable of observing the entire ecliptic. The extent of the WFD has evolved over time (as shown in Figure 2 and later discussed in Section 4.1.2), but no matter what the proposed variations to the WFD sky coverage are, the full extent of the ecliptic plane will not be incorporated into the WFD footprint. The NES minisurvey aims to remedy this situation by ensuring that higher-airmass observations of the northern ecliptic are taken as part of the LSST (LSST Science Collaboration et al. 2009, 2017; Schwamb et al. 2018b; Ivezić et al. 2019; Bianco et al. 2022). The NES region, shown in Figure 3, is composed of ∼604 pointings covering in total ∼5800 deg² spanning from 0° decl. to +10° ecliptic latitude. In order to make this goal achievable with the non-WFD time, the NES minisurvey has typically been implemented in the cadence simulations to receive a smaller number of visits per field (∼250; as shown in Figure 2) compared to the WFD. The NES minisurvey includes observations taken in a combination of the griz filters only, where solar system objects are typically the brightest. The observing time dedicated to the NES is explored in Section 4.1.3; here we focus on the impact of including or excluding the NES minisurvey from the LSST.

Zoom In Zoom Out Reset image size

The NES minisurvey is crucial for inventorying the outer solar system. About half of the ecliptic plane is covered within the WFD footprint. Over the 10 yr span of the LSST, inner solar system populations like the MBAs will complete full orbits. This means that MBAs in the northern hemisphere at the start of the survey will be in favorable positions to be detected within the WFD during the later years of the survey. This is not true for outer solar system objects whose orbital periods are well beyond ∼10 yr. Outer solar system bodies will only have a small fraction of their orbital periods covered by the LSST. Thus, the vast majority of TNOs located in the NES at the start of the survey will remain in the northern hemisphere, missing the WFD footprint. This is reflected in Figure 4, where the first two simulations plotted are baseline_v1.5_10yrs, which includes the NES minisurvey, and filterdist_indx2_v1.5_10yrs, which excludes the NES. TNO discoveries suffer nearly a 30% loss with the exclusion of the NES minisurvey, while there is only a very small drop for the inner solar system populations. Although not simulated at the time in this cadence experiment, populations that are more uniformly distributed on the sky (such as OCCs and ISOs) also benefit from the inclusion of the NES, which provides additional sky coverage and therefore more chances for discovery.

Zoom In Zoom Out Reset image size

Figure 4 also shows that the light-curve metrics for small MBAs suffer a bit more than a 15% loss when the NES minisurvey is not executed. Discovery relies on the object being above the 5σ limiting magnitude on three nights, but to perform light-curve inversion requires many more observations. The NES provides additional opportunities to observe those faint objects close to the LSST limiting magnitude, giving additional chances for the small MBAs to be observed in conditions where they might have sufficient S/N to contribute to shape modeling. The opposite effect is observed for the small PHAs and NEOs, which benefit in simulations without the NES minisurvey (about a 6%–10% increase in the light-curve inversion metrics). These objects are typically detected close to Earth and so quickly become too faint to be detected. Thus, pushing the time used for the NES minisurvey into additional WFD visits enables more observations where these small PHAs and NEOs are detectable and can have light curves measured. The opposite effect can be seen for the larger PHAs and NEOs. The larger PHAs and NEOs suffer a ∼10% drop in the light-curve metrics when the NES fields are excluded. Because large PHAs/NEOs are more likely to be above the limiting magnitude in an LSST image, surveying the NES creates new opportunities to monitor the brightness of large PHAs/NEOs as they pass by Earth. The Jupiter Trojans also take a significant hit when the NES is not included. This is likely due to their constrained positions on the sky.

Not captured in the MAF metrics are the benefits that the NES minisurvey provides to small body populations that are distributed asymmetrically across the sky. They would only be partially sampled without the NES observations, as discussed in Schwamb et al. (2018b). Two such cases are the Neptune Trojans and the resonant TNO populations. Over the 10 yr period, the vast majority of the leading Neptune Trojan L4 cloud is accessible only via observations of the NES as shown in Figure 5. Lin et al. (2019) find evidence for potential differences in the color distributions of the L4 (leading) and L5 (trailing) Neptune Trojans. The WFD and NES minisurvey combined are capable of sampling both clouds with sufficiently large numbers to test this further. Including the NES region in the LSST footprint enables characterization of the libration islands for the various mean motion resonances (MMRs) with Neptune (Schwamb et al. 2018b). Only observing half the ecliptic with just the WFD and Galactic plane minisurvey would impact the study of the resonant TNO populations, which preferentially come to perihelion at certain locations on the sky (e.g., Gladman et al. 2012; Gladman & Volk 2021).

Zoom In Zoom Out Reset image size

The NES minisurvey is crucially important for searching for additional distant planets in the solar system and testing the apparent orbital clustering of Sedna-like inner Oort Cloud objects (IOCs; q > 50 au and a > 250 au) and extreme TNOs (ETNOs; objects on orbits with q > 42 au and a > 150 au). It has been proposed that a giant planet ("Planet Nine") is gravitationally shepherding the distant planetesimals onto similar orbits with aligned orbital poles and longitudes of perihelion (Trujillo & Sheppard 2014; Batygin & Brown 2016; Sheppard & Trujillo 2016; Batygin et al. 2019; Brown & Batygin 2019, 2021; Oldroyd & Trujillo 2021). Recent modeling by Brown & Batygin (2021) combined with constraints from the Zwicky Transient Facility (ZTF; Brown & Batygin 2022) and the Dark Energy Survey (DES; Belyakov et al. 2022; Bernardinelli et al. 2022) predict Planet Nine to be residing at a semimajor axis of 700 au or higher. Although the current predictions made available in Brown (2022) do have Planet Nine distributed over a wide range of ecliptic longitudes, the most likely location of Planet Nine is close to the region where the Galactic plane intersects the northern ecliptic (see Figure 6). The bulk of the predicted Planet Nine sky locations are within the LSST footprint as implemented in the baseline_v2.1_10yrs simulation, which includes the NES minisurvey. Figure 7 presents the estimated on-sky V-band apparent magnitude distribution for Planet Nine from Brown (2022) and the predicted LSST r-band limiting magnitudes from the baseline_v2.1_10yrs run. Over a wide range of possible V − r colors, Planet Nine could potentially be bright enough to be visible with LSST images. If Planet Nine is bright enough to be imaged in single LSST exposures, the length of the survey in combination with the repeated coverage of the survey footprint effectively eliminates the possibility of missing Planet Nine in high Galactic latitude fields owing to coincidental overlap with another source. Closer to the Galactic plane, stellar crowding will be significant and identifying sources will be difficult. This may require community-optimized search algorithms to look for Planet Nine in these observations. Rubin Observatory is also exploring additional options to enhance source extraction near the Galactic plane (see Bosch et al. 2019). If current searches fail to find Planet Nine, Rubin Observatory will put the best observational constraints on the existence of Planet Nine over the next decade and will be the facility with the best chance of directly imaging it (Trilling et al. 2018b). As noted in Section 2.3.4, Rubin Observatory's SSP pipelines are only sensitive to moving objects at heliocentric distances ≲100–150 au. We fully expect that there will be several community-led efforts to find very slow moving distant objects in the LSST transient catalogs to search for Planet Nine and explore the IOCs and ETNOs. Therefore, it is still important to consider this science case for LSST footprint considerations.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Even if Planet Nine is not visible in the LSST images, the LSST would potentially be able to reveal its presence if the orbital alignment holds with the increased LSST sample of ETNOs and IOCs and matches the Planet Nine predictions. Whether or not the Planet Nine theory is correct, the distant IOCs and ETNOs are an important probe for studying the origin and evolution of the very distant outer solar system and testing alternatives to the Planet Nine theory (Morbidelli & Levison 2004; Brasser et al. 2006, 2012; Gladman & Chan 2006; Kaib et al. 2011; Zderic & Madigan 2020; Emel'yanenko 2022; Huang et al. 2022). Observing across the ecliptic will be crucial for creating a large enough sample to alleviate the challenging observational biases currently dealt with when combining the multiple data sets previously used to identify and test the apparent orbital clustering (Brown & Batygin 2016, 2019; Shankman et al. 2017; Bernardinelli et al. 2020; Napier et al. 2021).

The NES minisurvey (as described in Section 4.1.1) was proposed when the northern limit of the WFD footprint was initially set to be +2° decl. (see the baseline_retrofoot simulation in Figure 2). The originally planned WFD sky coverage used a simple cut in Galactic coordinates to identify the boundary of the WFD with the Galactic plane/bulge observing region (LSST Science Collaboration et al. 2017; Ivezić et al. 2019; Jones et al. 2020). Combining this boundary with the sky coverage requirements and visit constraints for the WFD set the original northern decl. limit. What sky is included within the WFD is a changeable LSST survey parameter, as long as the SRD requirements for the WFD survey area are met: at least 18,000 deg² with a median of 825 visits per field (Ivezić & the LSST Science Collaboration 2013). For extragalactic science, including cosmology and galaxy studies, and galactic science, such as the study of the Milky Way's structure, there have been proposals from the community requesting more of the WFD to be shifted to low-extinction and less crowded sky (Lochner et al. 2018, 2022; Olsen et al. 2018). Other arguments have also been raised for shifting the WFD footprint further northward, including overlap with future DESI (Dark Energy Spectroscopic Instrument; Abareshi et al. 2022) and Nancy Grace Roman Space Telescope observations (Olsen et al. 2018; Capak et al. 2019b). As this is a zero-sum game, visits are taken from near the Galactic plane with high stellar crowding and dust extinction and redistributed northward above 0° decl. This results in a fraction of the WFD survey now covering the NES region as seen in the Baseline v2.0 and v2.1 footprints shown in Figure 2. The SCOC has made the recommendation to use this new footprint as shown in Figure 2 extending the WFD northward, although the final decl. limits of the WFD and the exact boundary of the Galactic/high dust extinction region can still be fine-tuned (Ivezić & the SCOC 2021). As implemented in baseline_v2.0_10yrs simulation, the revised WFD footprint has two decl. boundaries spanning from −72° to +12° decl. with an interstellar dust extinction cutoff at approximately E(B − V) = 0.2 mag or A(V) = 0.6 mag (Ivezić & the SCOC 2021; Yoachim 2022), where E(B − V) is the dust reddening in magnitudes and A(V) is the total V-band extinction. The northern boundary of the WFD varies with R.A. in this revised northward footprint; this is partly due to other additional constraints with the scheduler. We note that baseline_v2.1_10yrs simulation uses the same footprint as v2.0 but incorporates the Virgo Cluster (α = 12 hr, δ = +12°) into the WFD (Yoachim 2022).

Expanding the WFD footprint northward will cover part of the NES for "free" with the time charged to the WFD time allocation, but part of the redistributed pointings in the 2°−12° decl. band are at high ecliptic latitude because part of the ecliptic plane crosses the Galactic plane in the southern hemisphere. Transferring WFD visits from the Galactic bulge region will reduce the number of photometric data points available for generating rotational light curves for some MBAs within the bulge, but how significant the impact is will depend on the exact shape of the footprint. The OCCs, NEOs, and PHAs are distributed across a wide range of ecliptic latitudes, so observations at higher ecliptic latitudes will still find small bodies in these populations. The same arguments that hold for outer solar system objects in Section 4.1.1 also apply in this case. Assuming a 2025 February 14 start date, Neptune's on-sky position will have changed by about 1 hr in R.A. and 8° in decl. by the end of LSST observations. Objects beyond 30 au will be moving slower than Neptune. Most of the TNOs and IOCs located in the NES at the start of the survey will remain in the NES throughout the duration of the LSST. As these distant objects do not move very far on-sky during the 10 yr survey, any observations of the NES are beneficial for discovery as long as not too much time is taken away from near ecliptic pointings in the southern hemisphere.

In Figure 8, we evaluate the impact of the new northward WFD sky coverage, comparing baseline_v2.0_10yrs and baseline_retrofoot_v2.0_10yrs simulations. All simulations predating the v2.0 simulations start from a variation of the WFD with the old +2° decl. limit. The v2.0 baseline also has additional changes to the observing cadence, including tweaks to the rolling cadence implemented and the exposure time for u-band visits. For an apples-to-apples comparison, the v2.0 release includes baseline_retrofoot_v2.0_10yrs, which uses the original v1.5–1.7 baseline WFD+NES footprint, leaving the other cadence parameters the same as the v2.0 simulation. The northward WFD footprint produces a slight increase in discovery metrics (all less than 5% change) for all populations except for the large Jupiter Trojans. There are also slight improvements in the light-curve metrics, with the smallest MBAs seeing more than a 10% increase with the extended WFD footprint. These increases may be more significant than represented in the MAF metrics if stellar crowding was taken into account. Although this comparison is to the v2.0 baseline, it will still hold true for the v2.1 baseline (baseline_v2.1_10yrs) that goes slightly more northward. We note that the addition of the Virgo Cluster is a minuscule change in area, and there are negligible impacts to any of the solar system metrics compared to baseline_v2.0_10yrs (as discussed in Section 4.7.2).

Zoom In Zoom Out Reset image size

For completeness, we briefly discuss the footprint experiments performed in the v1.5 and v1.7 releases that led to the revised northward WFD incorporated into the v2.0 and onward LSST cadence simulations. The discovery and light-curve metrics are shown in Figures 4 and 9. The v1.5 footprint simulations are set up with different overheads than the v1.7/2.0/2.1 WFD footprint simulations that allow for a larger number of visits to be distributed across the sky. The v1.5 footprint family uses these additional visits to explore the impact of typically adding more northern visits in various configurations, modifying the number of visits in the Galactic plane and NES (sometimes in differing filters), and some changes to the extent of the WFD footprint. The sky map showing the total numbers of visits per pointing in these v1.5 simulations is shown in Figure 10. In general, adding visits northward enhances TNO discovery statistics, and in most cases, there are only small impacts on the ability to obtain light curves and produce shape inversion models of inner solar system objects. Instead of adding a small number of visits, the v1.7 WFD footprint experiments explore WFD variations on a dust-extinction-limited footprint with variable north/south decl. limits. The total numbers of visits in these v1.7 WFD footprint experiments are shown in Figure 11. Overall, TNOs and outer solar system discoveries benefit the most, with the inner solar system object discoveries taking only a few percent loss in discoveries. The light-curve metrics for the most part see 5%–10% boosts in the various configurations of the more dust-free WFD, but they start to decrease more significantly for the smaller-sized MBAs, NEOs, and PHAs, as less of the ecliptic that intersects with the Galactic plane in the southern hemisphere is included in the WFD and the number of visits to those regions drops. Some caution needs to be taken in interpreting this result, as this loss may be less than what is shown by the metrics as detection efficiency of solar system objects (and by extension the ability to measure their light curves) decreases in crowded fields. The Jupiter Trojans are constrained in set locations on the sky; with small numbers of detections in the simulations, this is likely contributing to the variation observed in the light-curve metrics. Overall, these v1.5 and 1.7 footprint experiments show that moving visits northward is an improvement and paved the way for the optimized v2.0/v2.1 WFD footprint and full LSST footprint.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Three sets of simulations were done in which the visits to the NES and the Galactic plane were varied relative to the WFD coverage. In these simulations, varying numbers of extra visits to the NES or the Galactic plane are added at the expense of removing that observing time from the WFD. The simulations discussed below show how covering these specific areas with visits ranging from 1% to 100% of the WFD cadence affect metrics for solar system populations.

The vary_NES family of simulations included coverage of the fields in the NES at 1% of the WFD level, at 5%–55% of the WFD level in 5% increments, and at 75% and 100% of the WFD level; the baseline simulation has the NES fields at 30% of the WFD. The top panel of Figure 12 shows the discovery metrics for various solar system populations for these simulations. At low coverage levels for the NES (<10%), the discovery metrics for the TNO populations are reduced by more than 5% relative to baseline, and most populations show increasing discoveries as the NES coverage increases. Figure 13 shows the fraction of each solar system population (relative to baseline) that is observed in at least four of the grizy filters as a function of the NES coverage. If the NES is covered at ≲15% of the WFD, the fraction of TNOs (down to H = 6) that are observed in at least four filters is reduced by more than 20% compared to baseline; the NES must cover at least 25% of the WFD to not reduce this metric by more than 5%. The TNO populations are the most affected in both discovery and color light-curve metrics when the NES is not covered to at least 25% of the WFD level because they move slowly on-sky compared to closer-in solar system populations. Most of them will not move enough over the 10 yr LSST time span to move from NES fields to WFD fields. Covering the NES at <25% of the WFD also significantly decreases the number of faint MBAs and faint Jupiter Trojans that are expected to have light-curve measurements (bottom panel of Figure 12); all the populations generally improve in both the color light-curve and light-curve metrics as NES coverage increases.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The vary_GP family of simulations included coverage of the fields in the Galactic plane at 1% of the WFD level, at 5%–55% of the WFD level in 5% increments, and at 75% and 100% of the WFD level. Figure 14 shows the resulting solar system metrics for discovery and light-curve inversion. The discovery metrics for different solar system populations are all within 5% of baseline for these simulations, and the fraction of each population that has observations in multiple filters is also relatively unaffected. However, when the Galactic plane is covered at >30% of the WFD level, the fractions of faint MBAs, Jupiter Trojans, and PHAs with light-curve inversions all drop by 5% or more (increasing losses with increasing Galactic plane coverage) compared to the baseline simulations. This is likely simply a result of shifting time away from the WFD fields, decreasing the odds that the fainter solar system objects are above detection thresholds multiple times in the reduced number of visits to their fields.

Zoom In Zoom Out Reset image size

The plane_priority family of simulations varies how different regions of the Galactic plane are covered based on a priority map of the Galactic plane from the Rubin Observatory LSST Stars, Milky Way, and Local Volume and Transients and Variable Stars science collaborations. Some of these simulations also have pencil beam fields in areas of the Galactic plane that the WFD is not planned to cover. These targeted pencil beam fields would be visited at the same level as the WFD. The Galactic plane plane_priority simulations were completed with and without pencil beam fields, and two additional simulations were done with just four larger or 20 smaller Galactic plane pencil beam fields (the pencil_fs simulations). The discovery metrics for different solar system populations are almost all within 5% of baseline for these simulations, with only faint MBAs and faint Jupiter Trojans dropping slightly below those thresholds for the priority threshold at 0.1–0.2 (top panel of Figure 15). The color light-curve metrics (see Figure 16) are also generally similar across the family of simulations, with losses in the fraction of faint populations observed in four of the grizy filters that hover around 5% for the simulations with higher threshold values (above ∼0.4), with or without pencil beams; the simulations with the lowest thresholds show an enhancement in the color light-curve metric. However, for this entire family of simulations, the fractions of faint MBAs, Jupiter Trojans, NEOs, and PHAs with light-curve inversions all suffer >5% losses compared to the baseline simulation (bottom panel of Figure 15); again, this is likely due to additional time shifted away from the WFD fields. The set of simulations that cover the priority map at >0.6–1.2 threshold with or without pencil beams generally keep the light-curve inversion losses for these populations to between 10% and 20% compared to baseline. The simulations with four larger or 20 smaller Galactic plane pencil beam fields added in addition to the plane priority maps have worse light-curve inversion metrics for all solar system populations than simulations with just the priority maps.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The LSST cadence is currently planned with two exposures of equal length dubbed "snaps," nominally 15 s each, to be taken back-to-back at each visit to an on-sky pointing, except in the case of u-band observing (see Section 4.2.2). The original plan was for the Rubin Observatory data management pipelines to compare the two snaps in order to identify and flag pixel-level artifacts (e.g., cosmic rays). Source detection would be performed on the image resulting from coadding the two exposures (Ivezić & the LSST Science Collaboration 2013; Ivezić et al. 2019). We note that the Rubin SSP pipelines' discovery algorithm is agnostic to the number of snaps per visit, as it uses the transient sources detected in the coadded snaps image as input (Myers et al. 2013; Jurić et al. 2020).

There now exist many algorithms published in the literature for identifying cosmic rays in single astronomical images (e.g., Rhoads 2000; van Dokkum 2001; Shamir 2005; McCully et al. 2018). If these algorithms work well on LSSTCam images, there may be no strong reason for taking two snaps at each visit. The decision on whether to implement one snap or two snaps per visit will be made during commissioning of the LSSTCam and the Rubin data management pipelines (Ivezić et al. 2019) when the feasibility of single exposure cosmic ray rejection can be tested and the impact from satellite constellation streaks can be properly assessed (see Section 5.4). With two snaps, each planned visit has two camera readouts and two camera shutter openings and closings. Although the readout time and the movement of the camera shutter are relatively quick (well less than a minute), the summed time lost to these overheads over the entire 10 yr survey is nonnegligible in the case of the two snaps observing cadence. As can be seen in Table 4 for the v1.7 family of simulations (the most recent LSST cadence simulations exploring the number of snaps), switching to one snap generates an ∼8% gain in on-sky visits and an increase in the sky coverage reaching the WFD goal of 825 visits per pointing.

The impact of switching to a single-snap cadence will depend on what the added exposures are used for. The SCOC has not yet made any decision about snaps and how to partition out the extra visits. If a significant portion of the gained visits can be distributed across the entire LSST footprint or WFD and NES, the increase in exposures will add sky coverage and/or temporal coverage that will help with detection and monitoring of small bodies while enabling other science. In the v1.7 simulations, the extra visits gained in the one snap case were divided out evenly between the WFD and other parts of the simulation's survey footprint. This produces an increase in both the detection and the light-curve metrics (see Figure 17). The detection metrics for the small size end increase by a few percent. The extra visits provide additional chances for those objects near the survey brightness limit to get above the image 5σ limiting magnitude and be detected. The largest bodies see only a very slight increase because the majority of times when they land within an exposure they are already brighter than the limiting magnitude. The largest enhancement is seen with the light-curve inversion metrics, especially for the small end of the size distribution, where we find a >20% boost across the MBAs, PHAs, NEOs, and Jupiter Trojans. Like the case for discovery, the extra observations provide more opportunities for better temporal coverage to probe rotations and perform shape inversion. Since only six detections are required for discovery, it is the light-curve inversion metric that shows the true benefits for color and light-curve measurements from the extra on-sky observing.

Zoom In Zoom Out Reset image size

As noted in Schwamb et al. (2018b), there is some extra information that is potentially gained with two snaps per visit. Although the SSP pipelines are not currently planned to use any information from the individual snaps, bespoke community software could be developed to take advantage of the two exposures per visit. For those small body populations moving fast enough to be significantly trailed in the LSST images, such as NEOs and PHAs, the sequential snaps allow for (1) the on-sky direction of motion to be measured from the two streaks and (2) brightness variations (on the order of seconds) to be extracted from the streaks for ultrafast rotators. Only a very tiny fraction of the asteroids discovered will be rotating fast enough that sub-30 s resolution will be useful (Pravec & Harris 2000; Masiero et al. 2009; Warner et al. 2009, 2021; Hergenrother & Whiteley 2011; Chang et al. 2014, 2019, 2021, 2022). These are very limited benefits compared to the gains to all solar system populations from an extra 8% of on-sky observing time. Therefore, we recommend incorporating single-snap visits into the LSST cadence, if feasible.

The long_uX and u_long simulations explore the impact of using a single longer u-band exposure versus two shorter (15 s) exposures in the baseline. This was investigated since, because of the low level of sky background in u band, readout noise has a larger impact than in redder bands, and a single longer exposure allows us to significantly improve the u-band depth (see, e.g., Jones et al. 2020). The u_long family varies the duration of the single u-band exposure (30, 40, 50, or 60 s). It is expected that longer u-band exposures (40 or 50 s) will be advantageous for the detection of faint activity around solar system objects. Since the u band also encompasses emission from the CN radical around 388 nm (and to a lesser extent from the NH radical), there might be slight gains for active comets inside 3 au, increasing with decreasing heliocentric distances, but this has not been modeled in detail yet. However, longer u-band exposure times (starting marginally with 50 s but more strongly for 60 s) result in a lower number of observations being performed in other filters and thus decrease the number of faint Jupiter Trojans and PHAs detected, as well as the number of faint objects for which we can perform light-curve inversion, as illustrated in Figure 18. The long_uX uses a 50 s exposure, either keeping the same number of visits (long_u1) or reducing it (long_u2). Both of these tend to be worse than the baseline for solar system objects in terms of light-curve inversion in particular for faint objects, as illustrated in Figures 19 and 18. This results from the fact that light-curve inversion requires a certain number of observations above a certain S/N threshold, which might not be met for some objects in bluer filters, where most solar system bodies are fainter. The long_u2 family performs better for both detection and light-curve inversion metrics and was identified as a good compromise as long as it is not done together with any of the bluer_indxXX options mentioned in Section 4.3, which is shifting more visits to blue filters over redder filters.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The visits in the v ≥ 2.0 survey simulations are typically set to 2 × 15 s exposures in the grizy filters, while the u band has 1 × 30 s exposures. A series of simulations (v2.1 shave) has been run to explore the impact of different exposure times on the survey metrics compared to the family's baseline simulation. As seen in the top panel of Figure 20, the relative effect on the discovery rate of TNOs, faint OCCs, and faint MBAs diminishes significantly with shorter exposure times compared to the baseline exposure time configuration, as the 5σ limiting magnitude decreases with exposure time. Shorter exposure times have a greater effect on fainter absolute magnitude TNOs, dropping the discovery metrics by more than ∼5% for TNOs with 6 < H < 8 compared to TNOs with H < 6. The effect is similar for OCCs, with the discovery metrics decreasing by ∼5% for OCCs with 8 < H < 12 compared to OCCs with H < 8. The discovery of NEOs and PHAs does see a small improvement with the shorter exposure cadences owing to increased sky coverage resulting from the shorter exposure times allowing for more exposures to be taken (e.g., Jedicke et al. 2016). This improvement in the discovery metrics is greater for fainter PHA and NEOs with 16 < H < 22 than for bodies with H < 16. The enhanced discovery caused by the greater coverage is due to the fact that these closer-in objects tend to be detected at viewing geometries when they are closer to Earth and are thus brighter to compensate for their smaller size. Examples of faint, close-in asteroids whose discoveries are favored by the shorter cadence when they approach Earth include asteroids on Earth-similar orbits and meteoroids (e.g., Kwiatkowski et al. 2009; Granvik et al. 2012; Bolin et al. 2014, 2020a; Jedicke et al. 2018; Shober et al. 2019; de la Fuente Marcos & de la Fuente Marcos 2020b, 2022; Fedorets et al. 2020b; Naidu et al. 2021).

Zoom In Zoom Out Reset image size

When exposure times are increased to more than 30 s, more distant objects have improved discoveries, but the discovery of PHAs and NEOs diminishes. The decreased discovery of PHAs could be due to the decreased sky coverage in the longer-exposure scheme compared to the shorter-exposure scheme. The degradation in the number of PHAs and NEOs in the longer exposures could also be due to trailing losses from their higher rate of motion (e.g., Shao et al. 2014; Ye et al. 2019). One additional factor to consider in the longer exposure times is that they will be more susceptible to images being compromised from satellite trails, which is more likely in longer exposures (Tyson et al. 2020); see Section 5.4 for a detailed discussion.

The effect of shortening the exposure time improves the light-curve inversion metrics for all dynamical groups of objects included in the v2.1 cadence simulations as seen in the bottom panel of Figure 20. Shorter exposure times generally improve the light-curve inversion metrics owing to the improved coverage and improved density of detections enabled by shorter exposure cadences. The magnitude of improvement varies by dynamical class. For faint PHAs and NEOs with H = 19, the light-curve metric is almost doubled, with 20 and 22 s exposures compared to the baseline cadence. Larger PHAs and NEOs with H = 16 see a moderate improvement as well with the shorter exposures. The higher density of coverage will also be useful for the study of the rotation states of Jupiter Trojans and asteroid family members in the main belt (Hanuš 2018), e.g., as shown by the increase in the light-curve metrics for Trojans and MBAs. The benefits of wider and more frequent coverage of the sky to light-curve inversion may also extend to the monitoring and detection of activity within the asteroid belt (e.g., Moreno et al. 2017). The improvement for more close-in objects may be explained by their higher sky-plane motion, placing this in a wider range of possible areas of sky positions that is more easily covered with a shorter cadence. A good compromise exposure time for obtaining favorable discoveries for inner and outer solar system objects, as well as dense light curves, seems to be the 30 s exposure cadence.

An additional simulation, vary_expt_v2.0_10yrs, was designed to test the results of varying the exposure times between 20 and 100 s in the ugrizy filters to provide consistency in the image depth in different filters. As seen in the top panel of Figure 19, varying the exposure time between 20 and 100 s results in poorer discovery metrics relative to the baseline simulation for all classes of solar system objects used in the simulations. This is due to the fact that the longer exposures result in an overall decrease in survey coverage and a decreased chance to detect moving objects. As seen in the bottom panel of Figure 19, the effect on light-curve inversion metric is also worse for all classes of solar system objects. Therefore, varying the exposure times to achieve uniform visit depth is not recommended.

Nightly pairs in combinations of the g, r, and i filters are the most conducive to finding solar system objects. We explore the intranight separations in these combinations of filters. As discussed in Section 2.3.4, the SSP pipelines require that motion be seen between the two exposures (Myers et al. 2013; Jurić et al. 2020). If a solar system body has not moved sufficiently for it to be identified as a new transient source in the next visit, SSP will not be able to spot that moving object. The separation between nightly pairs sets the farthest distance at which SSP can detect moving sources. The time between repeat visits also directly impacts the number of pairs observed per night and thus the total area searchable for solar system objects. We aim to find the best pair time separation that increases the distances that SSP is sensitive to without making a significant trade-off in observing efficiency. This would allow the Rubin SSP to detect more distant TNOs while not compromising the discovery and characterization of the more inward solar system populations. We note that the tunable parameter here is the Rubin scheduler's goal for spacing the repeat visits in a given night. In reality, this will be a distribution centered about the ideal value the Rubin scheduler is aiming for. This is shown in Figure 24 for three examples from the v1.7 pair_times simulations that take mixed filters with ideal separations between 11, 22, 33, 44, and 55 minutes.

Zoom In Zoom Out Reset image size

Solar system objects appear to move fastest on-sky when they are at opposition, where the apparent motion is dominated by the parallax induced by Earth's movement. The on-sky rate of motion at opposition for a body exterior to Earth's orbit on a circular and coplanar orbit can be defined as

$$\begin{eqnarray}&&\frac{d\theta }{{dt}}=148\left(\frac{1-{r}_{h}^{-0.5}}{{r}_{h}-1}\right),\end{eqnarray} \tag{ 3 }$$

where r _h is the body's heliocentric distance in astronomical units and $\tfrac{d\theta }{{dt}}$ is the apparent motion at opposition in arcseconds per hour (Luu & Jewitt 1988). We assume 140 mas as a conservative estimate for the astrometric uncertainty for sources near the LSST detection limit and a 3σ positional shift for the SSP pipelines to successfully identify the moving object as a new source in the second observation (M. Jurić 2022, private communication). This translates to solar system bodies having to move at least 05 between the visits in order to become detectable by SSP, setting a minimum speed limit. In Figure 25, we estimate SSP's motion limit for the range of pair separations, including those explored in the pair_times simulations. The solid line represents the opposition on-sky rate of motion as calculated from Equation (3).

Zoom In Zoom Out Reset image size

The bulk of the classical Kuiper Belt extends from ∼42 to 47.7 au, the 2:1 MMR with Neptune, but the Kuiper Belt's scattered/scattering disk and detached/high-perihelion TNO population (with perihelion at ∼50–80 au) do extend well beyond that (Trujillo & Brown 2001; Petit et al. 2011; Adams et al. 2014; Bannister et al. 2018; Bernardinelli et al. 2022). Separations longer than 18 minutes are needed to search for objects beyond 80 au. Separations longer than 33 minutes start to slightly negatively affect the discovery metrics and significantly enhance the light-curve metrics, as plotted in Figure 26. The loss of discovery at fainter absolute magnitudes is less than 5% even at 55-minute spacings. As the time gap gets longer, the pairs are more vulnerable to interruptions, mostly from weather. The fraction of gri pairs peaks at 22 minutes, but the total visits and the on-sky area reaching 850 visits both increase with longer pair separations. The light-curve inversion metrics go up with longer gaps between intranight visits, due to the increase in the total number of visits, with a larger number of singleton images that are spread out across the observable sky (see Table 5). Having the Rubin scheduler aim for the two visits to be separated by 33 minutes is the best compromise between optimizing the number of nightly pairs completed and heliocentric distance probed. We note that the SCOC moved the LSST baseline strategy from aiming for 22-minute nightly pair separations to 33-minute ones from the v2.0 simulations onward (Ivezić & the SCOC 2021).

Zoom In Zoom Out Reset image size

The v2.0 long_gaps_np (long gaps, no pairs) simulations extend the time between nightly repeat exposures to 2–7 hr in various combinations. These simulations begin the extended pair separation either in Year 1 (delayed-1) or after Year 5 (delayed1827). This simulation family explores the impact of executing the long gaps strategy every night and less frequently, where the nightsoff parameter (the number of sequential nights with no long gap sequences) is varied. On the nights when the long gap observing is not active, the scheduler aims for 33-minute nightly pair simulations like the v2.0 baseline survey. These simulations are one option explored to potentially better capture fast-evolving astrophysical phenomena, as suggested by Bellm et al. (2022); additional strategies for addressing the temporal coverage of fast transients are explored in Section 4.4.4. As seen in Table 5, the fraction of gri pairs is largest when the Rubin scheduler is tasked with 33-minute pair spacings. Therefore, these hours-long separations are not going to be efficient in generating nightly pairs conducive for the moving object search. This can be seen in the discovery and light-curve metrics displayed in Figure 27. Across all populations, the light-curve metrics and detections decrease. The increased sensitivity to objects beyond 150 au is not worth the trade-off purely from a planetary astronomy perspective, but the simulations that do not have the long gaps in intranight visits occurring every night have less impact. The less time devoted to the large time gap pair observing, the less severe the hit to the discovery and light-curve metrics. Nonetheless, 33-minute pair separations are better optimized for outer solar system discoveries and the completion of repeat visits within the night.

Zoom In Zoom Out Reset image size

In the v1.5 simulations, cases were run with nightly pairs of visits performed in either matching filters (baseline_samefilt_v1.5_10yrs) or in mixed filters (baseline_v1.5_10yrs). The discovery metrics for solar system populations are largely unaffected by the choice (top panel of Figure 28). This is because the mixed-filter pairs in the cadence simulations contain filter pairs such as g − r and r − i, where the colors of solar system objects allow detections in both filters (we note that r − i pairings are better than g − r pairings for the reddest objects like TNOs). Similarly to what is shown in Figure 29, the color light-curve metrics for different populations are not significantly affected by the choice of same or mixed filters; there is likely an advantage to having mixed filters within the same night in that it could provide a single-night color estimate for objects that rotate slowly compared to the visit separation. For faint NEOs and MBAs, nightly pairs in the same filter do boost the light-curve inversion metric by 15%–20% (bottom panel of Figure 28). This is likely due to nightly pairs of faint objects that are only detectable in a small number of filters. However, faint Jupiter Trojans suffer a 30% loss in the light-curve metric for the same-filter pairs. This is likely related to the light-curve metric requirement that observations in a filter span at least 90° in ecliptic longitude. The Jupiter Trojans move more slowly on-sky compared to the other populations this metric is calculated for, and having same-filter nightly pairs reduces the number of different nights (and thus different longitudes) an object might be observed in that filter; for faint Jupiter Trojans, this appears to reduce the odds that successful detections in a given filter span the required range of longitudes.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In the baseline cadence (baseline_v2.1_10yrs), up to 20% of the pointings are visited more than twice per night. By adding an additional basis function to suppress these repeat visits to the Rubin scheduler algorithm, the additional visits can be distributed to different nights, thus changing the internight cadence or season length for a given field. The suppress repeats (no_repeat_rpw) family of simulations (as shown in Figure 30) explores these changes by considering six different values for the weight of the suppression factor, indicated as rpw, namely 1, 2, 5, 10, 20, and 100. This number basically reflects how strongly the suppress-revisits basis function influences the scheduler: the higher the number, the lower the number of revisits per night will be. Note that some regions of the sky will still be observed more than two times within a night if they are included in overlapping pointings.

An immediate consequence of redistributing the visits over different nights is a decreased total area with more than 825 visits per pointings, from a negligible effect (≲0.1% at rpw = 1) to a more significant effect of ≲5% at rpw = 20. However, because of the extended timeline, the discovery metrics are generally improved with respect to the WFD: a suppressing factor between 2 and 10 will increase the discovery rate for all the different families, while for rpw = 1, 20, or 100, there is only a marginal decrease (≲0.005%) in the discovery rate of faint TNOs and bright comets from the Oort Cloud. A suppressing factor equal to or larger than 10 will also impact the metrics of light-curve inversion, reducing up to ≈10% the number of faint MBAs and Jupiter Trojans for which inversion will be feasible. Summarizing, the "suppress visit" family cadence produces negligible effects on solar system science, with a marginal improvement on the discovery rates for rpw = 2 and 5 and a marginal decrement of the number of faint objects for which we will be able to perform light-curve inversion for rpw = 10, 20, or 100.

There is a strong desire among other Rubin Observatory LSST Science Collaborations to add a third visit in a different filter to aid in capturing and identifying fast (<1 day) transients by adding more color information (see Bianco et al. 2019) to the base survey of nightly pairs of visits that are separated by ∼20–30 minutes. The presto_gap family of simulations explores the effects of adding a third visit to the night's visits after a time period of 1.5–4 hr. Within the presto family of simulations, there are two significant subfamilies. The presto_half simulations explore the effect of adding the third image/triplet every other night rather than every night of the cadence, while the presto_gap_mix has a wider separation and difference in colors between the initial pair and the third visit (e.g., g + i , r + z , i + y rather than g + r, r + i, i + z initial pairs).

The main impact of adding this third visit is to dramatically decrease the amount of well-covered survey area (see Figure 31). This would have a large negative impact on science cases where the objects are sparse on the sky such as discovering rare objects (e.g., ISOs) or the onset of activity on solar system objects. The other large effect of adding the third visit is seen in the solar system object detection and light-curve inversion metrics and illustrated in Figure 32. Although there is some improvement in the detection of the brighter solar system objects at the shorter gap lengths in the 1.5–2.0 hr regime (see, e.g., the presto_gap1.5_mix simulation in Figure 32), this is not a high priority for the large-aperture capabilities of Rubin Observatory. For the vast majority of the other simulations and solar system populations, this family of simulations produces a 20%–75% decrease in the light-curve inversion metrics, well beyond our threshold for flagging these simulation families as bad for solar system science. The impacts are less dramatic for the presto_half subfamily, as might be expected, since the third visit is only carried out 50% of the time. The impact of the _mix version of a simulation (with the wider spread of observed colors in the third visit) is always worse than the corresponding "nonmixed" simulation run.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

As an alternative to the presto families discussed above, the long_gaps_nightsoffN family (not to be confused with the long_gaps_np family of simulations considered in Section 4.4.1) also adds a third visit in the same filter as one of the pairs (like the presto family). However, unlike the presto families, (1) the third visit forming the triplet is in one of the same filters as the earlier pair, (2) it only occurs if the first pair is in the griz filters, and (3) it occurs after a longer 2–7 hr gap from the initial pair than the standard ∼33-minute gap. This is done every N nights (N = 0...7), for example, long_gaps_nightsoff7 has the long gaps every seven nights, and long_gaps_nightsoff0 has "zero nights off" and the long gap third visit/triplets are done every night. These families additionally come in two flavors, delayed-1 and delayed1827, where the third visit/triplets start either immediately before the start of the survey (night −1) or in survey year 5 (night 1827), respectively.

Overall this family of simulations has much smaller detrimental effects (<10%) on the area covered (final third of Figure 31) and most solar system metrics, except for where this is done every or almost every night (the _nightsoff0 and _nightsoff1 simulations), which hit the area and light-curve inversion metrics hard (20%–60%; see final third of Figure 32). These families of survey strategy simulations, with longer gaps for the potential third visit in the night, could constitute a path toward satisfying the desires of other science goals without unduly compromising solar system science. The impact of the loss of sky area covered in all of these third visit simulations on the detectability of rare but high-value targets that are sparse on the sky, such as ISOs or very distant extreme TNOs (ETNOs and IOCs), needs additional simulations with these populations added. The addition of the OCCs to the later versions of the simulations, which are much more numerous on the sky than either ISOs or Sedna-like objects, and the corresponding drop in OCC discovery when adding the third visit show the downside of adding the third visit on discovering the rarer solar system populations.

The time balancing of DDF and WFD observations within the LSST has a noticeable impact on overall solar system detections and light-curve inversion capability. Simulations with a larger portion of survey time for DDFs were previously trailed in the v1.6 sims (ddf_heavy_) but were rejected, as they produced significant negative impacts on all solar system populations and their metrics, as well as failing to meet some of the key science requirements for the WFD.

The v2.0 simulations are the latest set of simulations that explore varying the fraction of total survey time allocated to DDFs. They test a more conservative variation of ±3% survey time spent on DDFs from the baseline value of 5%. In these simulations, any extra observing time is evenly distributed across the remaining components of the LSST. Both options are satisfactory for the discovery and light-curve inversion of solar system objects, for most or all populations (Figure 38). The simulation with 8% of time allocated to the DDFs (ddf_frac_ddf_per1.6 ⁴⁷ ) provides slightly worse results for discovery and light-curve inversion metrics compared to the simulation with 3% survey time (ddf_frac_ddf_per0.6 ⁴⁸ ), as WFD revisits are particularly important for light-curve infill and for linking the motion of solar system objects. However, both options are still within a negligible loss margin (<5%) on both metrics when compared to the baseline.

From the field population sensitivities, the key aspect of solar system science interest is the choice of rolling cadence for the COSMOS field and how it affects the small body metrics. The suggested types of DDF rolling cadences explored in the v2.1 cadence simulations are unique to the DDFs: in contrast to the "stripes" of the WFD discussed in Section 4.5, these simulations instead alter the frequency of observation for each individual DDF and the relative weighting of time between DDFs. They include cases where specific DDFs are observed only in certain years (e.g., only in the first 3 yr of LSST). Between v2.1 and v2.2, a large number of DDF strategy variations were considered, but in general solar system metrics have not been produced for these runs. The impacts of variations of DDF strategy while keeping the overall envelope of allocated DDF time and field location approximately constant are expected to produce negligible changes. Once a narrower range of DDF strategies are under consideration, solar system metrics will be produced and checked for potential impacts. Therefore, we consider how DDF "rolling" cadences would probably affect solar system science, with a specific focus on the highest-yield COSMOS DDF.

The Jupiter Trojans will complete approximately one full orbit during the span of the LSST. The slightly asymmetric populations lead and trail the giant planet in its orbit by ∼60°, with a mean libration amplitude of 33° from the center of their respective Trojan clouds (Marzari et al. 2002). The more populous L4 cloud's inclination distribution is centered around the ecliptic latitude of COSMOS, while the flatter L5 inclination distribution still encompasses COSMOS (Slyusarev & Belskaya 2014). The broad Jupiter Trojan libration distribution produces an on-sky distribution that has wide wings of consistent density around the libration centers. These orbital properties mean that Jupiter Trojans will be visible in the COSMOS DDF during distinct several-hundred-day observation periods within LSST (Figure 39). This will permit smaller-diameter Jupiter Trojans to be discovered than can be achieved by the WFD. Therefore, it is essential that the COSMOS DDF is observed at times when the Jupiter Trojans are passing through the field.

Zoom In Zoom Out Reset image size

Throughout the LSST, the COSMOS and other DDFs will provide a constantly refreshing sample of shift-and-stack-detectable TNOs smaller than can be seen in single frames of the WFD. As TNOs move slowly (<5'' hr⁻¹ for r _h > 30 au; see Figure 25), they will remain in the sidereally static DDF pointings for extended periods of time, longer than other solar system populations. For the LSSTCam FOV, the time in years for a TNO to pass through the field is given by

$$\begin{eqnarray}&&t=\displaystyle \frac{3\buildrel{\circ}\over{.} 5}{\alpha \,\tfrac{24\,\mathrm{hr}\,{\mathrm{day}}^{-1}\times 365.25\,\mathrm{day}\,{\mathrm{yr}}^{-1}}{3600^{\prime\prime} /^\circ }}=\displaystyle \frac{3.5}{2.435\alpha }\approx \displaystyle \frac{1.437}{\alpha },\end{eqnarray} \tag{ 4 }$$

where α is the opposition on-sky rate of motion in arcseconds per hour. For distances of 40–60 au, a TNO traverses the field in ∼5–8 months. In comparison, more distant TNOs (r _h ≥ 200 au) remain in the field for ≥2 yr. This means that the population of r _h ∼ 30 au TNOs observed in a DDF is refreshed ∼30 times during LSST as a result of (primarily Earth's) orbital motion, compared to r _h = 300 au TNOs, which would take a third of the full survey to pass through the field. COSMOS and, at lower yield, the other DDFs thus provide multimonth TNO orbital arcs that would determine parameters r _h and i to a precision useful for population studies. However, these arcs are generally too short to reduce uncertainties on a and e to levels sufficient for Neptune resonance classification (Volk et al. 2016). Deep revisits by LSST around the DDFs in later years to recover the DDF-sourced TNOs would be necessary for this additional improvement for outer solar system science. Therefore, TNO science is flexible relative to the DDF "rolling cadence" decision, as long as COSMOS and other DDFs are visited for approximately 2 yr at some point within LSST.

The twi_neo family of simulations use 50% of the available observing time during morning and evening twilight to perform a microsurvey of the low-SE (40° ≲ SE ≲ 60°) sky, which would otherwise not be observed during the WFD observing cadence (see Figure 40). The opportunity for LSST to observe the low-SE sky during twilight is the only time when viewing solar system objects inward to Earth is possible. Although surveys similar in nature have been carried out in the past, Rubin Observatory's large aperture size would put it in a unique position to provide a more sensitive search for several populations of solar system objects such as IEOs, Earth Trojans, and sungrazing comets than has been performed previously (Seaman et al. 2018). NEOs in the region of the solar system interior to Earth's orbit (including Atiras with orbits interior to the orbit of Earth and "'Ayló'chaxnims with orbits interior to Venus" orbit ⁴⁹ ) are the least constrained portion of currently available NEO models owing to observational limitations of objects at low SE (Greenstreet et al. 2012; Granvik et al. 2018). In addition, recent observational evidence and dynamical studies suggest that there are possible metastable regions in the innermost portion of the solar system where more objects on orbits similar to that of 'Ayló'chaxnim may be lurking and awaiting discovery (de la Fuente Marcos & de la Fuente Marcos 2020a; Greenstreet 2020; Popescu et al. 2020; Bolin et al. 2022, 2023; Ip et al. 2022; Sheppard et al. 2022).

Zoom In Zoom Out Reset image size

In addition to the discovery of IEOs, a LSST low-SE twilight microsurvey could enhance the discovery of ISOs; ISO 2I/Borisov was discovered during twilight by an amateur astronomer in 2019 (Borisov 2019). Routine observations at low SE could also provide prediscovery images of ISOs, enabling additional astrometric measurements for improved orbit determination of the often short-lived visitors (Bolin et al. 2020b; Ye et al. 2020). Low-SE observations of LSST-discovered ISOs would provide further opportunities beyond what the WFD observing cadence would offer for observing possible mass shedding, outbursting, or breakup events of these interstellar interlopers, as well as extend the amount of time these short-lived visitors can be observed. Monitoring the sky in the near-Sun region could also provide the opportunity to observe cometary outbursting or breakup events as non-interstellar near-Sun comets reach heliocentric distances <1 au, which may otherwise not be characterized. Observing comets (with origins from either within our solar system or interstellar space) as they reach the near-Sun region will better inform us of how insolation can process cometary surfaces and connect the near-Sun comet population to comets as a whole (Seaman et al. 2018).

Discoveries of PHAs can also be enhanced with the inclusion of a low-SE twilight microsurvey, improving our knowledge of and increasing warning times for potential asteroid impacts. In addition, the possible discovery of more Earth Trojans, which librate at the Earth–Sun L4 and L5 Lagrange points, would improve our knowledge of planetary impactor sources for both recent and ancient cratering events on Earth and the Moon (Seaman et al. 2018; Malhotra 2019; Markwardt et al. 2020). Earth Trojans also make attractive spacecraft mission targets owing to their low relative velocity with Earth. Lastly, observing asteroids in the near-Sun region with LSST can provide the opportunity to probe mechanisms responsible for the supercatastrophic disruption of asteroids with small perihelia (closest orbital distance to the Sun; Granvik et al. 2016) and test the extent to which this phenomenon occurs for asteroids that reach very small solar distances.

Due to the large amount of science for a wide variety of small body populations that would be made possible with a low-SE twilight microsurvey, a family of runs executing a variety of low-SE observing cadences during twilight have been included in the last few rounds of cadence simulations, the most recent of which are the v2.2 simulations. The v2.2 family is split into twi_neo and twi_neo_brightest, which execute the minisurvey when the Sun is above −178 and −14° elevation, respectively. The twi_neo and twi_neo_brightest simulations consist of 15 s exposures per visit and explore a variety of repeat visits, filters, and nightly twilight on-off cadences. The twi_neo_repeatX_Y_npZ_v2.2_10yrs and twi_neo_brightest_repeatX_Y_npZ_v2.2_10yrs families consist of X repeat visits (i.e., triplets or quads; all separated by ∼3 minutes) in Y filter(s) per pointing per twilight observed where Z = 1 (on every night), 2 (one night on/one night off), 3 (one night on/two nights off), 4 (one night on/three nights off), 5 (four nights on/four nights off), 6 (three nights on/four nights off), 7 (two nights on/four nights off). Figures 41 and 42 show the impact of these low-SE twilight microsurvey cadence options on the discovery and light-curve inversion solar system MAF metrics. We note that Figure 41 provides discovery completeness values that are not normalized to the baseline simulation's output since there are no 'Ayló'chaxnims discovered with the baseline survey cadence; this is the only figure that shows the outcomes of the MAF metrics analysis that is not normalized to the baseline simulation's output.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The discovery completeness for the 'Ayló'chaxnim population (NEOs with orbits interior to the orbit of Venus) for a variety of cadence options for this microsurvey are compared to that for the baseline survey (with no low-SE twilight microsurvey) in Figure 41. Each of the microsurvey cadence options provides a (sometimes much) higher discovery completeness than the baseline survey, which does not include any 'Ayló'chaxnim discoveries since 'Ayló'chaxnims are only visible at SEs smaller than the WFD cadence reaches. In the top panel of Figure 41, which uses the Rubin SSP discovery criteria of three nightly pairs in 14 days, the highest 'Ayló'chaxnim discovery completeness is reached when the low-SE twilight microsurvey is run every night with three repeat visits per pointing in either iz or riz filters (i.e., twi_neo_repeat3_iz_np1_v2.2_10yrs and twi_neo_repeat3_riz_np1_v2.2_10yrs). These microsurvey cadences would result in the completeness of the H ≤ 20.5 and the H ≤ 16.0 'Ayló'chaxnims increasing to ≈1.5% and ≈2%, respectively, providing the potential for a significant increase in the discovery of inner-Venus asteroids. Running the microsurvey with either three or four repeat visits during either the brightest twilight time (i.e., when the Sun is above −14° elevation) or full twilight time (i.e., when the Sun is above −178 elevation) produces similar results in 'Ayló'chaxnim discovery completeness across the various nightly "on"/"off" cadences. The largest discovery completeness increases for these options occur when the microsurvey is run every night using either iz or riz filters, which result in increases of ≈1% to ≈1.5%. Simply using the z filter does not get as large of a discovery boost as using either iz or riz filters. Unsurprisingly, the less often the microsurvey is run (fewer number of "on" nights), the lower the 'Ayló'chaxnim discovery completeness drops.

If, unlike the Rubin SSP requirement of three nightly pairs in 14 days, four detections in a single night with four repeat visits per pointing are required, the discovery completeness improves further. This is more typical of observing cadences used for NEO discovery by current surveys (e.g., Gehrels & Jedicke 1996; Larson et al. 2003; Tonry et al. 2018) Such a cadence would require the development and implementation of code outside SSP, which is designed only to use image pairs to make tracklets. Under this alternative cadence, running the microsurvey every night during the brightest part of twilight in either iz or riz (i.e., twi_neo_brightest_repeat4_iz_np1_v2.2_10yrs and twi_neo_brightest_repeat4_riz_np1_v2.2_10yrs in the bottom panel of Figure 41), the discovery completeness increases to ≈8% and ≈9.5% for the H ≤ 20.5 and H ≤ 16.0 'Ayló'chaxnims, respectively. Given that the baseline 'Ayló'chaxnim discovery completeness is zero, a low-SE twilight microsurvey thus has the potential for a dramatic shift in the discovery of asteroids interior to the orbit of Venus.

In contrast, Figure 42 (top panel) shows that for nearly all other solar system small body populations, running the low-SE twilight microsurvey every night produces the largest drops (≈3.5%) in discovery completeness, in particular for the fainter objects. This is because the low-SE twilight microsurvey takes time away from the WFD observing that would otherwise be performed during those twilight hours. This produces a drop in the discovery completeness for faint objects that would otherwise be discovered at larger SEs; faint objects are also harder to see than brighter objects when looking near the Sun. This drop is also increased when the microsurvey observations are only made with the z filter. On the other hand, bright (H ≤ 16) NEOs and PHAs get the strongest discovery boost when the microsurvey is run every night since more of the easily visible objects are picked up in the additional sky coverage. Of all the twi_neo family simulations, the two best options for discovery completeness for all included small body populations are twi_neo_repeat3_riz_np6_v2.2_10yrs, which includes three repeat visits (i.e., triplets) per pointing in riz filters where the low-SE twilight microsurvey is run with a cadence of three nights on/four nights off, and twi_neo_repeat4_riz_np4_v2.2_10yrs, which includes four repeat visits (i.e., quads) per pointing in riz filters with a cadence of one night on/three nights off. In these simulations, the only population to see a drop in discovery completeness are the H ≤ 12 OCCs with a maximum perihelia of 20 au, which get a ∼0.5% discovery drop; all other populations either match the baseline or gain an increased discovery completeness (up to ∼3.5% and ∼2%, respectively).

When considering the ability to perform light-curve inversions for PHAs, NEOs, MBAs, and Jupiter Trojans (bottom panel of Figure 42), running the low-SE twilight microsurvey every night is completely detrimental, with up to 40% drops for H ≤ 15 Jupiter Trojans and 20% drops for H ≤ 18 MBAs compared to the baseline survey that does not include any low-SE twilight microsurvey. Of the microsurvey options discussed above, the only option that does not drop the fraction of objects for which light-curve inversions can be obtained by >5% from that of the baseline survey (the level considered acceptable) is twi_neo_repeat4_riz_np4_v2.2_10yrs, which includes four repeat visits (i.e., quads) per pointing in riz filters, where the low-SE twilight microsurvey is run with a cadence of one night on/three nights off. This simulation keeps light-curve inversion at baseline level or above with up to a 30% increase above baseline levels for H ≤ 16 PHAs. As described above, this option also performs well for discovery completeness, which provides baseline-level performance or higher (up to ≈2%) for all included small body populations except the H ≤ 12 OCCs with a maximum perihelia of 20 au, which again get a ∼0.5% discovery completeness drop. An additional benefit to having four repeat visits instead of three repeat visits is better resiliency to contamination by satellites streaks (for additional discussion, see Section 5.4).

In general, running the low-SE twilight microsurvey less frequently proves better for both discovery completeness and light-curve inversion when all solar system small body populations are considered. Furthermore, running the low-SE twilight microsurvey at an infrequent cadence boosts discovery completeness overall, including for the 'Ayló'chaxnims, which also see a significant discovery completeness enhancement in both discussed simulations (to ≈0.5%–0.75% for three repeat visits in riz with a cadence of three nights on/four nights off or ≈0.15%–0.25% for four repeat visits in riz with a cadence of one night on/three nights off). Light-curve inversions are also enhanced when the low-SE twilight microsurvey is run infrequently (once every 3 days with four repeat visits per pointing in riz filters) compared to the baseline survey cadence. Given these enhancements and the large amount of science for a wide variety of small body populations that would be made possible with a low-SE twilight microsurvey, we thus strongly encourage an infrequently run low-SE twilight microsurvey to be included in the LSST survey cadence from the start of the survey. We note for the reader that the significant increases shown here from the discovery metrics when the twilight low-SE microsurvey is included will not translate into the exact same gains in the actual LSST 'Ayló'chaxnim, H ≤ 16 NEO, and H ≤ 16 PHA discovery yields. It depends on the size and albedo distribution of these populations, which is not included in the calculation of our discovery metrics (see Section 2.3.1). The significant increase in our metrics does show that including the microsurvey will significantly enhance LSST's chances of finding new'Ayló'chaxnims and other IEOs, but the actual number of new discoveries may be very small.

Given the numerous scientific benefits and enhancements in solar system discoveries and light-curve inversions that will come from running the low-SE twilight microsurvey, as described above, we recommend avoiding waiting to start this microsurvey until Year 2 or later in the 10 yr survey. One additional reason for starting this microsurvey in Year 1 is the increasing number of satellite constellations being sent into low Earth orbit. These satellite constellations are most problematic for astronomic observations during twilight hours, when the numerous satellites are brightest in the sky (for further discussion of the impact of satellite constellations on solar system science, see Section 5.4). With the number of satellite constellations continuing to increase, and plans for that increase to continue for years to come, the problem of contamination will only get worse during the later years of the 10 yr LSST survey. We thus recommend starting the low-SE twilight microsurvey in Year 1 of operations in order to reduce the level of satellite contamination as much as possible and enable the most solar system science. The SCOC has recently made a recommendation for a low-SE NEO twilight microsurvey to be included in the survey strategy starting in Year 1 of LSST, with further opportunities to explore the final details of the implementation (Bianco & the SCOC 2022).

Finally, we look at possible further cadence enhancements and/or software improvements. All currently analyzed cadences assume either three or four repeat visits (i.e., triplets or quads), but the discovery criteria are the same as used for WFD observations (linking three tracklets over a 14-day period). Under such assumptions, the standard observing strategy of requiring pairs should be examined as well. Hints that it may perform better are in Figure 41: note how repeat3 cadences systematically produce more discoveries than repeat4; a hypothetical repeat2 cadence may even further increase our sensitivity to 'Ayló'chaxnims. We recommend that such cadences be simulated and analyzed.

Should the analysis conclude that three- or four-tracklet cadences are still preferred, we would recommend that the Rubin Observatory consider reporting such tracklets to the MPC immediately (within 24 hr). This would allow for third-party follow-up of such objects, which may be few enough and interesting enough that it is feasible that they may be followed within days of discovery and not require explicit self-follow-up by Rubin. It would then be interesting to examine this scenario in further detail (since it would not follow the traditional SSP detection method) and quantify the number and purity of high digest 2 tracklets (i.e., short-arc moving object detections likely to be solar system objects) Rubin would identify and report on a nightly basis, as well as the typical apparent magnitude of reported tracklets (i.e., assess whether the broader community's NEO follow-up system would be able to keep up with this modified reporting method).

A number of additional microsurveys requiring 0.3%–3% of overall survey time were submitted in response to the 2018 white paper call on survey strategy. ⁵⁰ These microsurveys include the following:

• 1.  
  Adding extra fields: virgo (adds the Virgo Cluster to WFD), carina (1 week yr⁻¹ in Carina Nebula), smc_movie (short g exposures in SMC for two nights), roman (covering the Roman bulge field twice per year).
• 2.  
  local_gal_bindx 〈I〉: Covers 12 Local Group galaxies with extra gri exposures.
• 3.  
  too_rate 〈X〉: Follows Targets of Opportunity, where X is the number per year,
• 4.  
  north_stripe: Adds northern extension stripe up to decl. +30 (illustrated in the final panel of Figure 2).
• 5.  
  short_exp: Takes up to 3 × 5 s exposures in Year 1 of the survey.
• 6.  
  multi_short: Takes sequences of 4 × 5 s exposures in each filter, with the aim of obtaining 12 total sequences of short exposures per year, achieving ∼700 exposures per pointing.

As shown in Figure 43, the effects of the majority of these microsurveys on the discovery and light-curve metrics that are of most concern for solar system science are minimal (≲5%) since this involves a very small fraction of the overall LSST survey time. The exception to these generally minimal effects is seen in the multi_short simulation (final column in Figure 43). This survey strategy causes a ∼5%–25% drop in the number of objects detected, particularly for the fainter solar system object populations (this outcome is to be expected with the switch of 12 of the ∼82 exposures per pointing per year to much shorter exposures). While the vast majority of the additional microsurveys have little to no effect on the metrics of most interest to solar system researchers, there remains the possibility that the combination of several of these microsurveys could "constructively interfere" in such a way as to cause a large impact later. However, these microsurveys also only impact a very small amount of the survey time and the overall survey strategy, and so the decision on the details of microsurveys may well also be delayed until later in the cadence decision process. A combined set of microsurveys will likely be the subject of further simulation runs later when the larger and more influential parts of the cadence strategy are decided on.

Zoom In Zoom Out Reset image size