RAMI1 phase
The purpose of RAMI experiments in direct mode is to assess whether different RT models can generate comparable spectral Bidirectional Reflectance Factors (BRF) given precisely defined environmental and observational conditions.
Two categories of such conditions were setup, for homogeneous and heterogeneous environments, respectively. Within each of these two categories of scenes, the scattering medium could contain non-dimensional scatterers (turbid case) or finite-sized scatterer (discrete case).
Graphical visualizations of these environments are available on their respective pages.
By following the above links instructions are obtained that allow for the implementation of the prescribed test cases.
Once a RT models has been set up to represent one of these environments, the spectral and directional reflectance fields can be generated (for precisely specified illumination and observation conditions), and further information can be derived, such as the albedo of the geophysical system, or, the fraction of absorbed radiation.
Prior to the performing of any RT model simulations, please refer to the 'definitions' pages for detailed instructions regarding the angular sign conventions for BRF simulations, as well as precise definitions of the required leaf normal distributions and other RT model technicalities. Also read the relevant file naming and formatting conventions that must be adhered to by all participants.
Two types of intercomparison experiments were proposed for direct mode simulations during the first phase of RAMI.
Conceptually these test cases consisted of structurally homogeneous and heterogeneous environments, to be represented both as a turbid medium and in a discrete manner where the location of every scatterer is explicitly described.
In both cases RT simulations had to be performed with scatterers having optical properties that are typical for the solar domain. In the homogeneous case, however, simulations with conservative scattering properties were also required.
Due to the various approaches of RT models to represent the structural properties of a given geophysical medium, and the radiation transfer therewithin, their results should be expected to differ somewhat from each other and it is not clear that one can be declared better than another on the basis of such comparisons.
The aim of this RAMI exercise was not to validate the models, but, rather, to assess the degree of coherence between them.
The scientific question then becomes: Can we distinguish between any two participating models on the basis of the reflectance fields they generate for a well defined scene? The details of the assessment methodology are described in the paper [Pinty et al. (2000)] that has appeared in the special section of the Journal of Geophysical Research issue with the proceedings of this conference. In essence, the primary criterion to quantify inter-model variability within the context of RAMI is a measure of distance between BRF fields generated under identical geophysical and geometrical conditions. This metric was computed as the sum of normalized differences between pairs of reflectances, where the sum is taken over all models, all wavelengths, all scenes, and all illumination zenith angles. The normalized differences themselves are the ratio of the absolute value of the pairs of reflectances, divided by their sum. This total sum is further divided by the number of cases considered. Similar metrics can be used to estimate model discrepancies for each geometric condition of illumination or observation, scene or wavelength. This methodology can lead to the identification of models which are significantly different from other models, but, as indicated above, a full assessment of their validity requires a more detailed analysis of their physical basis.