NPZ models as such are 0-dimension models and can't be used to present turbulent mixing. However, in most cases, they are used in conjunction with a hydrodynamic model which simulates eddy diffusion and water transport.

In that case and if assuming that all of the nirogen is a solute you can assume that the nirogen behaves like the salinity.

As for salinity - you can use the rules of sollution to assess the extent of mixing.

for instance if at T0 you have salinity of 35 at the surface and salinity of 32 at the deep water and at T1 you have salinity of 34 at the surface you can assume that the surface water are now a solute that contains 33% old deep water and 66% old surface water. where the 33% was calculated usend (Surface salinity delta)/(Vertical salinity delta at T0).

When actually making the calculation of diffusion effect, you will first have to take into consideration the effect of advection.