Dear community of modelers

The regression terms in statistical models sometimes be can derived from first-principle mechanistic models that have their own terms and are derived from a very different process.

The terminology in such link models can get confusing and I am wondering if there is some naming convention.

Here is a simple example of a paper of linked models (Anderson 2005, Ecological Modeling). The model describes fish survival in terms of mean free path length during fish migration. The paper links statistical and mechanistic models giving

(logS)2 = aX2 + bT2

where S is a dependent variable and X and T are independent variables or covariates. By normal convention, a and b are regression parameters that can be estimated with statistical methods. Linking the regression and mechanistic models the statistical parameters are explained in terms of the mechanistic model terms as

lambda= a-0.5 and omega = (b/a)0.5.

What should the mechanistic model terms be call? My thought is to call them "coefficients" to distinguish them from regression model "parameters"

This highlights the different classes of terms and the linkage of statistical and mechanistic models, which is not often done. Is this useful or simply pedantic?