Fig. 3. Four scenarios of identifiability for the SIAR model of Fig. 1.

All panels show the system dynamics (solid line) and the evolution of the system when the vector of parameters undergoes a variation δp such that ∥δp ∥ = 0.3 ∥ p∥ along the first eigenvector of M (dashed lines). (A) and (C) display configurations for which the observed variables (ι, r) are not sensitive to the variation, i.e., the model parameters are not identifiable, while (B) and (D) show the opposite case. Furthermore, (A) and (B) present scenarios for which the unobserved variables (s, a) are insensitive to the variation, meaning that they are predictable; vice-versa, (C) and (D) show the case in which the variables s and a are sensitive.