The DP0.3 Simulation#

The DP0.3 data set is composed of two sets of catalogs containing real and simulated solar system and interstellar objects: one that represents LSST results after one year, and one that represents LSST results at the end of the 10 year survey. The DP0.3 data set is hosted on the Rubin Science Platform (RSP) and available to all DP0 delegates.

Credit: The DP0.3 data set was generated by members of the Rubin Solar System Pipelines and Commissioning teams, with help from the LSST Solar System Science Collaboration, in particular: Pedro Bernardinelli, Jake Kurlander, Joachim Moeyens, Samuel Cornwall, Ari Heinze, Steph Merritt, Lynne Jones, Siegfried Eggl, Meg Schwamb, Grigori Fedorets, and Mario Juric.

Observing strategy#

The DP0.3 simulation uses the LSST baseline v3.0 cadence (see page 44 of the Survey Cadence Optimization Committee’s Phase 2 Recommendations). This observing strategy includes the North Ecliptic Spur and deep drilling fields (DDFs). Two of the DDFs are close to the celestial equator (have declinations near zero), and so are particularly beneficial for the detection of Solar System objects. Field revisit rates, which are driven by the needs of Solar System science, are on average 33 minutes apart. The baseline v3.0 observing strategy also contains a twilight survey for near-Earth objects (NEOs).

Simulated objects and detections#

The DP0.3 simulation includes hundreds of millions of detections of millions of objects, real and synthetic (see below), including trans-Neptunian objects (TNOs), main belt asteroids (MBAs), interstellar objects (ISOs), Hildas and Trojan asteroids, long-period comets, and near-Earth objects (NEOs).

The ObjectsInField package was used to generate an ephemeris from an object catalog and the Rubin cadence, from which source detection and measurement were simulated using the SurveySimPostProcessing package (Merritt et al. 2023, in prep).

The simulation includes astrometric scatter and photometric variations based on the objects’ color class (silicaceous and carbonaceous, see below), the exposure’s filter, and the object’s phase angle. However, rotation curves or complex geometry are not included. In other words, each DP0.3 object is a uniform, textured sphere in one of two colors. Any changes over time in an object’s apparent magnitude are due only to changes in its distance and phase angle.

Due to time constraints, DP0.3 does not contain u- or y-band detections. This decision was made in part because the majority of objects will have very low signal-to-noise ratio in u and y, and object discoverability is driven by the gri bands.

While neither the two-color nor the missing u/y band simplifications should prevent testing and exploration of the simulated Rubin data set, there is a plan to provide a more realistic simulation in an upcoming update.

Real objects from the MPC#

The DP0.3 simulation contains all objects in the Minor Planet Center Orbit (MPCORB) Database as of May 1 2023, except for the ~400 objects that have no absolute magnitudes. Out of these objects, Rubin detects 97% (1.2 million) of them in the simulated 10-year survey.

Synthetic object populations#

The DP0.3 simulation includes 91% of the objects in the Synthetic Solar System Model (S3M) catalog and 12,148 simulated ISOs, and two long-period comet populations from LSST SSSC Cadence Optimization, 5,000 with q < 5 au and 5,000 with q < 20 au. In the simulated 10-year survey, Rubin detects 24% (3.2 million) S3M objects, 20% (2,429) of the simulated ISOs, 68% (3,395) of the q < 5 au long-period comets and <1% (39) of the q< 20 au long-period comets.

Objects were simulated in two color classes: S and C (silicaceous and carbonaceous, see Veres for more details), with colors and slope parameters (GS) as shown in Table 1.

Table 1: The C and S color classes used for DP0.3.#

























Combining real and synthetic moving objects#

To combine the real and synthetic populations while maintaining S3M’s well-chosen orbital distributions, the Hybrid Solar System Catalogue Creator (Hybridcat) was used. Hybridcat removes the closest-matching synthetic object to each real object, creating a population with all of MPCORB and most of S3M that closely match S3M’s orbital distributions.

Truth data#

The following truth parameters per observation can be found in the DiaSource tables for both the 1-year and 10-year DP0.3 catalogs.

nameTrue: The true MPC name of the object that generated the detection (for real objects).

raTrue: The true right ascension for each observation.

decTrue: The true declination for each observation.

magTrueVband: The true V-band apparent magnitude for each observation. Use the color terms in Table 1 to recover the true apparent magnitude in the band of the observation.

Furthermore, the MPCORB tables contain injected rather than measured orbital parameters, so in this sense the MPCORB tables can be thought of as “truth tables”.

Known issues#

There are several known issues with the DP0.3 simulation.

Problems with the designation columns in the MPCORB table. The MPCORB table has two columns for the designation of a moving object, fullDesignation and mpcDesignation, and they each have known issues. First, the fullDesignation values appear to have a prefix of “2011” which should not be there. Second, the mpcDesignation column has not been “packed” as expected, but still has an arraysize parameter of 8. When values from this column are returned via a query, they are truncated versions of the full designation (without the 2011) with all characters after the eighth removed. The recommendation for now is to use the full designation and ignore the prefixed “2011”.

Relatedly, note that some objects which are purely simulated (and not clones of real objects) have designation formats that do not match real moving object designations (e.g., spaceships have prefix “ET”). These are not issues and not mistakes, but also do not necessarily represent future designations of real LSST-detected moving objects.

Underestimated number of detections. DP0.3 is known to underestimate the number of observations by up to 20 or 30%. This is a result of the combination of issues mentioned below.

Color classes. Only two color classes have been simulated, with no variance, which is not realistic. There is also an issue in that objects are about 0.2 mag fainter than they should be across the board, because of an error that set \(V-r = 0\). These too-faint magnitudes mean that source detection rates especially for outer solar system objects will be underestimated with DP0.3.

Missing u and y band. The u and y band detections were not simulated for solar system objects in DP0.3. As SS objects are typically quite faint in u and y, and there are fewer visits in these filters than in griz, this has only a small (if any) impact on moving object identification rates.

Overly dim detection magnitudes. There is an issue with trailed magnitudes that leads to overly dim detections (e.g., by up to 10 mag). This is an issue for only a few to ten thousand detections out of billions.

The camera footprint was slightly inaccurate. However, the size was approximately correct. This means that while the total number of detections are still representative of the future LSST, individual objects might not have been detected for a given visit when they should have been.

All ISOs are similar. They all have similar dates of perihelion passage (pass the Sun at similar times) and they do not have a distribution of absolute H magnitudes (they are all bright).

The number of ISO and LPC objects exceeds predictions. There are many more ISO and LPC objects in the simulation than the LSST is projected to observe, by orders of magnitude (only a few are expected). This is not exactly an issue: these populations were inflated on purpose so that a diversity of properties could be simulated.

Small biases in the reported PSF magnitudes. A small bias (roughly 0.02 mag) was identified in the slope parameter G (i.e., measured value - truth value of G) in all griz bands. Also, offsets between the intrinsic absolute magnitude in V band and recovered absolute magnitude in the LSST filters were found to be larger than listed in the filter-conversion table above (Table 1). Section 4 in the notebook tutorial DP03_04b describes these two biases. The DP0.3 simulation team found out that this is because the DP0.3 catalogs reported standard point-spread function (PSF) magnitudes without considering the apparent motion of solar system objecst. Moving objects appear in an image as trailed sources, resulting in underestimation of the source’s true flux as well as lower photometric signal-to-noise ratio (S/N) when conducting standard PSF photometry. Once object trailing is taken into account in photometry, these biases disappear and will be fixed in subsequent releases.