Figure 1
General structure of archNEMESIS for a forward model simulation. The input information is stored as the attributes of several classes that, together with the model parameterisations we may want to include, carry all the information we need to simulate the spectrum of a planetary atmosphere, which is calculated within the forward model.
Figure 2
Schematic of the two main types of model parameterisation base classes included in archNEMESIS. Models are implemented as specific instances of either PreRTModels, which modify the input to the radiative transfer routines, or PostRTModels, which modify the computed spectrum. Each instance must define three main methods that specify how the parameterisation interacts with the rest of the archNEMESIS reference classes.
Figure 3
General structure of the forward model. The attributes of the ForwardModel class, given by the reference classes, are first updated based on the model parameterisations. Using these updated attributes, the code calculates the atmospheric paths and performs the radiative transfer calculations, simulating the spectrum of the planetary atmosphere we want to model. The modelled spectrum is then convolved with the instrument function to simulate the specifics of the instrument we are modelling. Finally, the computed spectrum is updated by any relevant model parameterisations.
Figure 4
General structure of archNEMESIS for a retrieval run. Similar to the forward model scenario, the reference classes and model parameterisations provide all the necessary information to run the simulation. The retrieval engine runs the forward models and modifies the model parameters to find the set that produce a best fit between the measured and modelled spectra. The retrieval engine will provide as an output the best fit to the measurement, as well as the values of the model parameters that produce the best fit.
Figure 5
Example of comparison exercises between the DISORT and the archNEMESIS radiative transfer codes. The top three panels show the simulated spectrum from Mars under different scattering conditions and using pre-tabulated correlated-k tables with a full-width at half-maximum of 0.025 µm, while the bottom panels show the relative difference between both codes. The left panels show the comparison when no dust is present in the atmosphere. The middle panels show that when including a visible dust column optical depth of 0.5 under no scattering conditions, while the right ones show the effect of simulating the multiple scattering field.