Upcoming telescopes such as the James Webb Space Telescope (JWST), the European Extremely Large Telescope (E-ELT), the Thirty Meter Telescope (TMT) or the Giant Magellan Telescope (GMT) may soon be able to characterize, through transmission, emission or reflection spectroscopy, the atmospheres of rocky exoplanets orbiting nearby M dwarfs. One of the most promising candidates is the late M-dwarf system TRAPPIST-1, which has seven known transiting planets for which transit timing variation (TTV) measurements suggest that they are terrestrial in nature, with a possible enrichment in volatiles. Among these seven planets, TRAPPIST-1e seems to be the most promising candidate to have habitable surface conditions, receiving
M dwarfs are the most common type of stars in our galaxy, and rocky exoplanets orbiting M-dwarf stars will likely be the first to be characterized with upcoming astronomical facilities such as the James Webb Space Telescope (JWST). Ultra-cool dwarfs (
TRAPPIST-1 is an active M-dwarf star
Upstream of future JWST characterization of TRAPPIST-1e, it is important to derive constraints on its possible atmosphere to serve as a guideline for the observations. For this purpose, 3-D global climate models (GCMs) are the most advanced tools
Global climate models (GCMs) are 3-dimensional numerical models designed to represent physical processes at play in planetary atmospheres and surfaces. They are the most sophisticated way to model the atmospheres and oceans of real planets. GCMs can be seen as a complex network of 1-D time-marching climate models connected together through a dynamical core (see description below). Each 1-D column contains physical parameterizations for radiative transfer, convection, boundary layer processes, cloud macroscale and microscale physics, aerosols, precipitation, surface snow and sea ice accumulation, and other processes at varying levels of complexity.
The motivation behind this experimental protocol is to evaluate how some of the differences between the models can impact the assessment of the planet's habitability and its observables through transmission spectroscopy and thermal phase curves with upcoming observatories such as JWST. The intercomparison protocol was designed to evaluate three possible sources of differences between the models listed.
Note that a particular emphasis will be given to the differences of cloud properties between the models, because they may have a large impact on the strength of the spectral signatures simulated by current radiative transfer tools
Four GCMs (in their planetary version) are initially included with THAI:
the Laboratoire de Météorologie Dynamique – Generic model the Resolving Orbital and Climate Keys of Earth and Extraterrestrial Environments with Dynamics the Exoplanet Community Atmospheric Model (ExoCAM, available on GitHub: the Met Office Unified Model
By publishing our protocols in advance of the intercomparison work, we hope that other teams will also use this protocol to compare their own GCM with the four GCMs of this study.
TRAPPIST-1e is up to now one of the best habitable planet candidates for atmospheric characterization through transmission spectroscopy with the JWST. Therefore, it is also an obvious candidate for an experimental protocol for GCM intercomparison.
In Table
TRAPPIST-1 stellar spectrum and TRAPPIST-1e planetary parameters from
For THAI, we have chosen a set of four planetary configurations with increasing complexity. We have chosen to start with benchmark cases of dryland planets with
We have therefore two lands planets (Ben1 and Ben2) and two aqua planets (Hab1 and Hab2). Note that Ben1 and Hab1 share the same atmospheric composition of 1 bar of
Surface contours for surface temperature, thermal emitted radiation (top of atmosphere, TOA) and reflected stellar radiation (TOA) for Hab1 simulated by the four GCMs: the UK Met Office United Model (UM), the Laboratoire Météorologie Dynamique Generic model (LMDG), the Resolving Orbital and Climate Keys of Earth and Extraterrestrial Environments with Dynamics (ROCKE-3D) and the National Center for Atmospheric Research Community Atmosphere Model version 4 modified for exoplanets (ExoCAM).
In each case, it is crucial to start each simulation with the same initial conditions. The simplest approach is then to start with an isothermal atmosphere. For THAI, we fixed the initial surface and atmosphere temperature at 300 K. The atmospheric configurations for the two benchmark (dry land) cases and two habitable cases are listed in Table
In Fig.
THAI experimental protocol.
The surfaces considered in THAI (Table
Globally averaged surface temperature
The model spatial resolution is an important parameter because every process taking place at a subgrid level would be parameterized and those parameterizations often diverge between the models. Similarly, the model time steps control the numerical stability and accuracy. However, the choice of time steps is fundamental to how each model operates under a given parameterization, and arbitrarily fixing these parameters may prevent some models from correctly and fairly performing the intercomparison. In addition, models should be compared using the specifications that they commonly use for exoplanet studies. Therefore, for the sake of the THAI, we do not impose the model spatial resolution nor time steps. However, note that we recommend (but this is not a requirement) the radiative time step (a parameter much more flexible than the others among the models) to be set to 1800 s. This value should provide a good coupling of the radiation with temporal changes to the atmosphere without slowing down the model too much.
We also ask the contributing scientists to disable the subgrid gravity wave parameterizations in their model. Indeed, not all of the models have implemented a gravity wave parameterization, and some have prescribed or predicted gravity wave formation, tuned for Earth topography and meteorology. Therefore, to avoid differences in atmospheric dynamics, especially above the tropopause, we recommend to not include the subgrid gravity wave parameterizations in this intercomparison. Gravity waves whose wavelengths are greater than the model grid are explicitly resolved in the models and do not need to be modified.
Note that under the requirements of the protocol, the atmospheric simulation of TRAPPIST-1e may actually not represent what each individual model can simulate with all their parameterizations fully activated. This is especially true for the sea ice and snow albedo parameterization. Therefore, complementary to the Hab1 case, we propose the Hab1
Instantaneous fields to be output by the GCM. For each diagnostic, the mean value and the standard deviation are computed from data output at the specified frequency and number of orbits for the case. OLR and ASR correspond to outgoing longwave radiation (at TOA) and absorbed shortwave radiation (at TOA), respectively; SW and LW correspond to shortwave and longwave radiation, respectively; CF corresponds to cloud fraction and MMR corresponds to mass mixing ratio. The empty set symbol (
To compare the difference between models of a particular (instantaneous) output variable, both the average and standard deviation over the specified frequency and number of orbits for the case will be computed.
Four categories of outputs frequently used in climate simulations have been selected: radiation, surface, atmospheric profiles and clouds. The radiation outputs are the outgoing longwave radiation (OLR) and absorbed shortwave radiation (ASR) for clear and cloudy skies, also commonly known as emitted thermal and absorbed stellar fluxes, respectively, both at the top of the atmosphere (TOA). The surface outputs are the temperature map, the downward total SW flux and net LW flux, and the open ocean fraction (for Hab1/Hab1
All the simulations should have reached radiative equilibrium at TOA at
To facilitate comparison between each GCM, we ask the contributing scientists to provide their outputs in netCDF format. The contributing scientist will be able to upload their data to a public permanent repository at
The main objective of THAI is to highlight how differences in atmospheric profiles produced by each GCM are going to impact the predictions of atmosphere detectability and observational constraints for habitable planet targets such as TRAPPIST-1e
Note that while additional simulations with a simple Newton cooling model, a 1-D column model, or with cloud radiative effects disabled would help to better understand the differences due to the dynamical cores and cloud physics, they would also dramatically increase the computational time, amount of data and effort. THAI aims to be easily reproducible and not time consuming in order to reach many GCM user groups. The five simulations proposed in THAI should be enough to understand the main differences between the GCMs and their impact on the observables. THAI could also be used as a benchmark for future GCM intercomparisons that specifically aim to understand each differences between the models.
THAI is an intercomparison project of planetary GCMs focused on the exciting new habitable planet candidate, TRAPPIST-1e. Because rocky exoplanets in the habitable zone of nearby M dwarfs have the highest chance to be the first Earth-size exoplanets to be characterized with future observatories, TRAPPIST-1e is currently the best benchmark we could think of to compare the capability of planetary GCMs. In this first paper we have presented the planet and GCM parameters to be used in this experiment which already has four GCMs included (LMDG, ROCKE-3D, ExoCAM and UM), but we hope more GCMs will join the project. The results of the comparison of these four models will be given in a second paper and a THAI workshop is planned for fall 2020.
ExoCAM
TJF lead the THAI project and has written the article. ETW ran the simulation for Fig. 1 and plotted the figures. Every author contributed to the development of the THAI protocol, to the discussions and to the editing of the article.
The author declares that there is no conflict of interest.
Martin Turbet acknowledges the use of the computing resources of OCCIGEN (CINES, French national high-performance computing center).
The THAI GCM intercomparison team is grateful to Anong's Thai Cuisine restaurant in Laramie, WY, USA, for hosting its first meeting on 15 November 2017. We would like to thank the reviewer Daniel D. B. Koll, the anonymous reviewer and the topical editor Julia Hargreaves for comments that greatly improved our article.
This project has received funding from the GSFC Sellers Exoplanet Environments Collaboration (SEEC), which is funded by the NASA Planetary Science Division's Internal Scientist Funding Model. This project has also received funding from the European Union’s Horizon 2020 research and innovation program under the Marie Skłodowska-Curie actions (grant agreement no. 832738/ESCAPE). Michael J. Way and Anthony D. Del Genio acknowledge funding from the NASA Astrobiology Program through participation in the Nexus for Exoplanet System Science (NExSS).
This paper was edited by Julia Hargreaves and reviewed by Daniel D. B. Koll and one anonymous referee.