In the last blog , we discussed the importance of a global registry of parameters used in the Geosciences. Such parameters are effectively coordinates in models of physical phenomena, many of which may span multiple geosciences. It thus makes sense to expand our registry to include the models themselves, creating a registry of phenomenon models.
When we say phenomenon models, we are not necessarily talking about computational models, and our description need not allow us to compute anything. The phenomenon model is more like a schema that lists the global parameters involved in the model. You might think of a Monge expression, as in dimensional analysis, where we simply identify parameters: U, L, f, T, ρ (U = wind velocity, L = tornado diameter, f = coriolis frequency, T = temperature, ρ = air density) and write g(U, L, f, T, ρ) = 0, where the specifics of g are unknown.
The nature of the phenomenon may impact the names assigned to the parameters. For atmospheric phenomena, we would speak of wind velocity, whereas in the ocean, we may speak of ocean current velocity. However, in either case, they are velocities, and this must be taken into account in the parameter registry.
A phenomenon model registry must reference the parameters in the parameter registry that participate in any given phenomenon model (see my previous blog entry ). In addition, the model must provide a name (or names) and a text description, and it must have a globally unique identifier.
Relationships between phenomenon models are also critical. Some phenomena are interdependent and this can be captured, in part, through phenomenon model relationships. A particular relationship exists between models that are defined by what are effectively “equivalent” parameters, such as in the case of an experimental or measurement model and a theoretical model. For example: measurements have been taken of the ocean current velocities and scales, surface temperature, etc. In this case, a phenomenon model is constructed using parameters such as “Ocean_current_velocity_measured_from_satellite”, while in the second case (theoretical), we may have the same set of equivalent parameters (e.g. “ocean_current_velocity”). These relationships between phenomenon models, and the equivalences or other relationships between the parameters, must also be captured in the phenomenon model registry.
As in the case of parameters, data sets and software must be mapped to the entries in the phenomenon model registry. These mappings provide the description of the data set(s) and software semantics, and include, in particular, software for simulation as well as software that simply interpolates or “models” the measured data.
A phenomenon model registry extends further the construction of common semantics for the geosciences.