A weather report for the hot Jupiter HD 189733b II.

The weather forecast of HD 189733b is further analyzed by the LEAP team, this time led by Graham Lee. They report further cloud coverage and dynamics in their newly published paper in Astronomy and Astrophysics (Lee et al. 2017).  Learn more about their results from Graham’s summary below.

In part I of this series, we coupled a radiative-hydrodynamic (RHD) atmospheric model to a microphysical cloud formation module. We simulated the atmosphere of the hot Jupiter HD 189733b and found that the planet is likely to host a highly inhomogeneous cloud structure dependent on latitude, longitude and depth of the atmosphere. In our next paper in the series, we compare our results to available Hubble Space Telescope and Spitzer Space Telescope observations of the exoplanet. We do this by post-processing our large scale cloudy RHD results using the Monte Carlo radiative-transfer (MCRT) method.

Figure 1 Images of the simulated planet at 1 micron wavelengths and at a planetary phase of 60o (eastern terminator region). Left: Incident stellar scattered light. Right: Emitted thermal light. The cloud structures are more prominent in the left image and a twilight effect (scattering past the day-night terminator) is present. In the right image the hotter equatorial jet is very recognizable compared to the cooler higher latitude regions and the nightside of the planet. (Lee et al. 2017)

Monte Carlo radiative-transfer in a nutshell

A particular method for radiative-transfer is the Monte Carlo radiative-transfer (MCRT) technique. A MCRT simulation is a 3D stochastic, statistical sampling method of determining the local radiation field in an environment. We track the 3D random walk (or path) of millions of simulated “photon packets” through our 3D RHD simulation results. We do this by sampling a random number which govern the packets interactions with the surrounding medium. For example, if a photon packet undergoes a scattering event with a cloud particle, the new propagation direction of the photon packet is determined by sampling a random number and comparing that number to the “phase function”, which is the probability of the packet being scattered at any one angle (e.g. a random number of 0.1 could correspond to a new path slanted 10 degrees from the original direction). By simulating the path and interactions (scattering, absorption) of many millions of packets, we can count the total amount of energy escaping towards a certain direction and produce synthetic observable properties of our large scale atmospheric model. Figure 1 shows an example output from the MCRT simulations applied to our RHD model.

Figure 2  Albedo of our simulated planet compared to the observational data obtained by Evans et al. (2013). Our model is reasonably consistent with the B-band data but does not predict the drop in albedo at the V-band. Our simulation is able to decompose the effect of the different scattering material present in the atmosphere. Most scattering comes from cloud particles with a small H2 Rayleigh scattering component. (Lee et al. 2017)

Albedo/reflection spectra of HD 189733b simulation

When a transiting planet passes behind its host star, both the emitted light from the planetary atmosphere and reflected incident starlight disappear from the total luminosity of the star and planet system, resulting in a dip in observed flux of the system during the secondary eclipse of the transit. The albedo of a planet is defined as the fraction of incident starlight that is scattered back into space. For example, a planet with an albedo of 0.2 at a wavelength of 300 nm, returns 20 % of the incident star light at 300 nm back into space. The albedos of many hot Jupiter planets have been measured by the Kepler space telescope, and some by the Hubble Space Telescope. Figure 2 shows the albedo of our HD 189733b simulation produced from the MCRT simulation compared to the geometric albedo observations using the Hubble Space Telescope by Evans et al. (2013). We also combine our albedo and emission spectra to compare to dayside observations of HD 18973b from Hubble and Spitzer Space Telescopes (Fig. 3).

Figure 3  Dayside luminosity ratio of our simulated planet compared to HST and Spitzer measurements (Evans et al. 2013, Crouzet et al. 2014, Barstow et al. 2014 and Knutson et al. 2012). Our simulation compares well to the observed trends in the dayside emission spectra. (Lee et al. 2017)

We find that our model is generally consistent with the observations but has some offsets with some data points. However, we do not fit the “V-band” (Visible wavelength range) dip in albedo as observed by the HST measurement. This suggests the presence of an absorbing material present in the atmosphere of HD 189733b which was not accounted for in our modelling. This absorber has been noticed in other hot Jupiter planets and it is currently a mystery what causes these dips in albedo at these wavelengths. Overall though, this general agreement with the observations suggests that the cloud coverage and temperature structures of our large scale simulation are probably a reasonable expectation of the dayside properties of HD 189733b.

Figure 4  TESS and CHEOPS predicted dayside albedos and bandpasses (dashed line). CHEOPS has a larger albedo due to the effect of photons scattered by clouds at optical wavelengths, which TESS is not sensitive too. (Lee et al. 2017)

Predictions for TESS and CHEOPS

We also produce predictions of the geometric albedo and phase curve behaviour of HD 189733b for the upcoming NASA Transiting Exoplanet Space Survey (TESS) and ESA CHaracterising ExOPlanets Satellite (CHEOPS). Figure 4. shows our results shows our TESS and CHEOPS predictions, and includes a Kepler prediction (even though HD 189733b will not be observed by Kepler) for comparisons. We found that our simulated dayside atmosphere would be 10% brighter when viewed by the CHEOPS mission compared to the TESS instruments. This is due to the fact that CHEOPS is more sensitive to the optical scattering component (cloud cover), as TESS is more sensitive to infra-red thermal emission from the planet. By comparing the albedos and phase curves between the CHEOPS and TESS instruments, an idea of the extent of the cloud coverage on the observed planet may be inferred. This may be useful for helping decide which planets to observe with more time limited spectroscopic instruments such as the James Webb Space Telescope (JWST).

Figure 5  RGB channel GIF of the incident scattered light varying with planetary phase. The real planet is probably slightly bluer than the brownish colour presented here due to an unknown absorber at visible wavelengths (See Fig. 2; Lee et al. 2017).

RGB channel phase curves

Lastly, we produced a RGB channel phase curve output using the albedo results. This is provided in GIF form in Fig. 5. The real planet is likely to be slightly bluer than the reddish brown colour presented due to the unknown V-band reducing the R channel’s magnitudes.

For more details check out the original paper on ADS:

G. Lee, K. Wood, I. Dobbs-Dixon, A. Rice, Ch. Helling, 2017, A&A, 601, A22

The LEAP Group can be found here:


And finally, don’t forget to like us on Facebook: