WFI Direct Imaging Simulations in Support of Exoplanet Microlensing
Summary
The Roman Space Telescope microlensing survey is designed to discover cold, low-mass exoplanets with semimajor axes beyond about 1 AU through the microlensing effect (Spergel et al. 2015). The goal of the GULLS simulation (Penny et al. 2019) is to predict the yield of the Roman Space Telescope microlensing survey. For the nominal baseline Roman Space Telescope design, the overall prediction is for ∼1400 bound exoplanets with mass greater than 0.1 M⊕. Given in the tables below is the baseline design — WFIRST Cycle 7, including telescope, instrument and other parameters — assumed for the simulation. Note that the parameters given below were the ones used for the simulations, but they may not necessarily accurately reflect the most recent changes in the conceived Roman Space Telescope telescope, instrument and survey parameters.
Generating the Sky Scene
The simulation is performed using the GULLS code, which produces simulated images for each microlensing event it simulates. GULLS was first introduced and described in Penny et al. (2013), where it was called MaBμLS, and has then been renamed to disambiguate it from the www.mabuls.net online tools (Awiphan et al. 2016). GULLS uses a smooth background plus stars from the Besançon Galactic model (see below). At each epoch a new realization of the counts is made. Counts from stars, smooth backgrounds and instrumental backgrounds (thermal background and dark current) are Poisson realized, and fluctuations from readout noise are Gaussian realized. See the table below for the assumptions about telescope and instrument parameters used in the simulations. For point source simulation, GULLS uses well sampled numerical PSFs generated using the WebbPSF tool (Perrin et al. 2012) with parameters from Cycle 5. Then unweighted relative aperture photometry is performed in a three pixel by three pixel square. GULLS uses user-supplied functions to compute microlensing magnification, including any effects that the user wants to model (e.g., binary lens, parallax).
| Name | Value |
| Orbit | L2 |
| Mirror diameter (m) | 2.36 |
| Obscured fraction (area) | 13.9% |
| Detectors | 6 × 3 H4RG-10 |
| Plate scale (arc-sec/pix) | 0.11 |
| Field of view (deg2) | 0.282 |
| Baseline duration (years) | 5 |
| Name | Value |
| Total survey duration (d) | 432 |
| Seasons | 6 |
| Season duration (d) | 72 |
| Survey area (deg2) | 1.96 |
| Fields | 7 |
| Average slew and settle time (s) | 83.1 |
| Primary bandpass (μm) | 0.93–2.00 (W149) |
| Secondary bandpass (μm) | 0.76–0.98 (Z087) |
| Photometric precision | 0.01 mag @ W149 ∼ 21.15 mag |
| Name | W149 | Z087 |
| Zeropoint∗ (mag) | 27.615 | 26.387 |
| Exposure time (s) | 46.8 | 286 |
| Cadence | 15.16 min | 12.0 hr |
| Exposures per field | ∼ 41000 | ∼ 860 |
| Saturation (mag) | ∼ 14.8 | ∼ 13.9 |
| Bias (counts/pix) | 1000 | 1000 |
| Readout noise★ (counts/pix) | 12.12 | 12.12 |
| Thermal + dark† (counts/pix/s) | 1.072 | 0.130 |
| Sky background‡ (mag/arcsec2) | 21.48 | 21.55 |
| Sky background (counts/pix/s) | 3.43 | 1.04 |
| Error floor (mmag) | 1.0 | 1.0 |
| Saturation§ (103 counts/pix) | 679 | 679 |
Notes
∗ Magnitude that produces 1 count per second in the detector.
★ Effective readout noise after multiple non-destructive reads.
† Sum of thermal backgrounds (caused by infrared emission of the telescope and its support structures etc.) and dark current.
‡ Evaluated using a zodiacal light model at a season midpoint.
§ Effective saturation level after full exposure time. For the Roman Space Telescope it is assumed that thanks to multiple reads, useful data can be measured from pixels that saturate after two reads, so for a constant full-well depth, the saturation level increases with exposure time.
GULLS Yield Microlensing Simulator
GULLS uses a modified version of the Besançon Galactic model (Kerins et al. 2009, Robin et al. 2003), with source star counts normalized to HST star counts from Clarkson et al. (2008) and Calamida et al. (2015), and the microlensing event rates are normalized to corrected event rates (Sumi & Penny 2016) from the MOA-II survey (Sumi et al. 2013). Information on similar Euclid simulations is available in Penny et al. (2013) where the simulator was first discussed and referred to as MaBμLS, Manchester-Besançon microLensing Simulator, and the general information in that paper is applicable to Roman Space Telescope observations as well. GULLS has also been used for an analysis for the predictions for the detection and characterization of a population of free-floating planets with K2 Campaign 9 (Penny et al. 2017).
Update on the Galactic Model
The microlensing yield depends critically on the underlying Galactic model used within the simulation. This is an extremely active field of research. Recent analyses that make specific reference to the Roman microlensing survey include Terry et al. (2020) and Koshimoto et al. (2021).
