Assessment of a regional physical–biogeochemical stochastic ocean model. Part 2: Empirical consistency