LSST Simulations

The LSST Simulation Framework is a powerful set of tools that enables end-to-end simulations tracing the properties of the LSST system design from the underlying cosmology through to derived data products.  Each component of the framework provides a broad range of capabilities - from the generation of the statistical properties of a year's worth of observations to targeted simulations of stars to evaluate how well the pointspread-function can be interpolated across a sensor. 

Simulation software and data products are available to the LSST project and to the community. 

The LSST Systems Engineering group manages the LSST simulation framework which is depicted in this schematic illustrating the type of simulations and modeling tools that have been built and and the range of applications. The framework comprises three main components: a catalog simulator (CatSim) capable of returning catalogs of astrophysical sources (e.g. stars, galaxies, and solar system objects) with properties and noise characteristics that are representative of what the LSST will observe to its coadded depth; an image simulator (Phosim) that is capable of returning images with characteristics consistent with the design of the LSST (i.e. with astrometric, photometric and ellipticity distributions that are appropriate for a large, wide-field long exposure telescope); and an observing strategy simulator (OpSim) that can generate sequences of LSST observations (and their summary statistics) that meet the 10 year cadences and depths required by the survey (while accounting for the expected performance of the telescope and site).

The need for science simulations arises because the requirements described in the SRD are a simplification of a complex system that incorporates the physics of the universe, the performance of the subsystems, and our ability to analyze these data under varying conditions. Engineering simulations such as Zemax2 or FRED3 have been used to define the optical design of the system. While detailed, these modeling tools do not couple the astrophysical properties of the sky nor the changes in observing conditions to the system performance. They are not designed to scale to the size of simulations of the LSST universe with 20 million sources per focal plane image (to a coadded depth of i=26.8).

In contrast, science or system simulations provide the ability to take a value specified within the SRD, which incorporates opto-mechanical, atmospheric, electronic, and software components together with the underlying astrophysical distributions of sources and evaluate which systematic uncertainties are most sensitive to individual components (i.e. assuming we can model the simulation components at the appropriate level of fidelity). A science simulation framework can provide an end-to-end implementation of the full flow of photons and information to evaluate how well we can achieve the SRD requirements or a simplification of the flow of information to identify the subcomponents and their contribution to the overall performance.

For additional information, see the 2014 SPIE paper - "An end-to-end simulation framework for the LSST".