The reference state is a dry, isentropic atmospheric profile based on the surface potential temperature. Many atmospheric profiles, both realistic and idealized, can deviate significantly from this reference state at certain levels. Should we check whether the anelastic approximation is being violated?