The Great Oxidation Event was a period during which Earth’s atmospheric oxygen (O₂) concentrations increased from ~10⁻⁵ times its present atmospheric level (PAL) to near modern levels, marking the start of the Proterozoic geological eon 2.4 billion years ago. Using WACCM6, an Earth System Model, we simulate the atmosphere of Earth-analogue exoplanets with O₂ mixing ratios between 0.1% and 150% PAL. Using these simulations, we calculate the reflection/emission spectra over multiple orbits using the Planetary Spectrum Generator. We highlight how observer angle, albedo, chemistry, and clouds affect the simulated observations. We show that inter-annual climate variations, as well as short-term variations due to clouds, can be observed in our simulated atmospheres with a telescope concept such as LUVOIR or HabEx. Annual variability and seasonal variability can change the planet’s reflected flux (including the reflected flux of key spectral features such as O₂ and H₂O) by up to factors of 5 and 20, respectively, for the same planetary phase. This variability is best observed with a high-throughput coronagraph. For example, HabEx (4 m) with a starshade performs up to a factor of two times better than a LUVOIR B (6 m) style telescope. The variability and signal-to-noise ratio of some spectral features depends non-linearly on atmospheric O₂ concentration. This is caused by temperature and chemical column depth variations, as well as generally increased liquid and ice cloud content for atmospheres with O₂ concentrations of <1% PAL.

The netCDF (.nc files) data were produced from simulations using the WACCM6 (CESM2.1.3) model - see https://www2.acom.ucar.edu/gcm/waccm and https://doi.org/10.1029/2019JD030943. These simulations were performed on the ARC4 supercomputer at the University of Leeds, and on the Cheyenne supercomputer at the NCAR-Wyoming Supercomputing Center. 

The atmospheric data from each simulation were output in terms of 5-day instantaneous 'snapshots' (cam.h1 files). These file names are set by a naming convention. e.g. b.e21.BWma1850.f19_g17.baseline.Spectra.001.cam.h1.0013-01-01-00000.nc, where 'baseline' indicates the pre-industrial simulation, and b.e21.BWma1850.f19_g17.10pc_o2.Spectra.002.cam.h1.0033-02-05-00000.nc, where 10pc_o2 indicates the 10% the present atmospheric level of oxygen (O₂) simulation. In this example, the date is indicated at the end: 0033 is the year, the middle 02 is the month, and the final 05 is the day of that month. So b.e21.BWma1850.f19_g17.10pc_o2.Spectra.002.cam.h1.0033-01-01-00000.nc is the instantaneous data output at midnight (UTC) on the 5th of February, on the 33rd year of the simulation.

The Earth's ephemeris data was collected from the PSG website and placed into a .csv file.

The .txt files are produced by processing the WACCM6 output through a locally installed version of the Planetary Spectrum Generator (PSG; https://psg.gsfc.nasa.gov/). They are found in the *_PSG_output_clouds_*pc.zip files. The filenames correspond to the pre-industrial (PI), 150% PAL (One50), 10% PAL (Ten), 1% PAL (One), and Zero1 (0.1% PAL) simulations. The two numbers following the 'clouds_' give the particle sizes of water and ice clouds, and then 10 pc, 25 pc, and 50 pc indicate the distance to the system in parsecs. The year and date are also given in each file name.

To produce the figures in the manuscript, the Python programming language was used.

Python programming language. See the WACCM_to_PSG.py file for converting the WACCM data into .gcm binary files.