You are here : Home > Workpackages > WP3


Published on 26 February 2019



Development of retrieval techniques able to handle retrieval of multiple data sources. In addition, we perform the retrieval directly on the light-curves, instead of on the derived transmission/emission spectra. This allows us to systematically address asymmetries in the planet atmosphere and even more importantly take into account the effects of this on the error budget of the retrieved parameters. The new technique, developed in the framework of this workpackage, will have to deal with emission and transmission data simultaneously in a single model. This requires simultaneous, consistent modelling of different regions on the planet.

We will provide a homogenous set of retrieval parameters for all sources in the catalogue, allowing direct comparison. This allows recognising relative trends robustly.

The retrieval based parameter set will be used as starting point for providing ab-initio modelling of all sources.


Description of work

The work needed to be done to reach the objectives of this work-package can be broken down in two tasks; development of the retrieval technique and applying the retrieval to the sources in the catalogue.

The postdoc hired on this work-package will be stationed at SRON and closely collaborate with UCL. For task 1 the expertise at UCL will be especially crucial, requiring multiple longer work-visits.

Task 1: Development of retrieval technique

The retrieval technique needs to be adjusted to be able to simultaneously model transmission and emission observations. This requires simultaneous modelling of the day side of a planet and the day-night terminator. Also, chemical abundances will be different for the upper atmosphere (transmission) and the deeper layers (emission). A method has to be devices that takes this into account. We will study how to constrain the chemistry, possibly using equilibrium chemistry or parameterised disequilibrium chemistry. We now aim at retrieval based on Bayesian Nested Sampling analysis (building on the Tau-REx framework), but alternative solutions will be investigated.

Since exoplanet observations are highly challenging, data reduction is not straightforward. Therefore, it is generally a good idea to stay close to the data. Within this task, we will device a method that analyses light-curves rather than the corresponding spectra extracted from the light-curves. This has the advantage that correlations of the light-curve model fitting and systematic artefacts of spectral extraction can be combined with atmospheric model parameter retrievals and modelled in a coherent framework. In addition, it allows us to address the importance and influence of asymmetries in the planet atmosphere. Due to the coding complexities involved and the inter-disciplinary nature of the problem (one that is actively addressed by this proposal), a combined light-curve-atmospheric retrieval framework has not been attempted in the literature and is an entirely novel approach to a more global understanding of model and data degeneracies. This makes our error analysis of the retrieval method a much more robust framework.


Task 2: Applying the retrieval

Within WP2 a large catalogue of observations is created for a sample of objects. Within this task we will apply the developed retrieval technique to this dataset. We will provide a homogeneous set of exoplanet parameters and models. An example of atmospheric retrieval results is given in figure X. Since the parameters are retrieved in exactly the same way for all objects, the systematic errors, caused by e.g. modelling assumptions, are the same for all objects. This gives us the unique possibility to spot relative trends far more robustly and enable, for the first time, statistically coherent comparative planetology. The catalogue as well as all models will be made publically available on a dedicated web-server.