Incorrect units for tracer flux and tendency diagnostics

[aekiss]

It looks like non-temperature tracer fluxes and tendencies are given generic units that don't take into account the units of the tracer (ie it is assumed the tracer is dimensionless):

MOM5/src/mom5/ocean_tracers/ocean_tracer_advect.F90

Lines 870 to 929 in b68f137

	else
	id_tracer_advection(n) = register_diag_field ('ocean_model', trim(T_prog(n)%name)//'_advection', &
	Grd%tracer_axes(1:3), Time%model_time, 'rho*dzt*advection tendency', &
	'kg/(sec*m^2)', missing_value=missing_value, range=(/-1.e18,1.e18/))
	id_tracer_advection_on_nrho(n) = register_diag_field ('ocean_model', trim(T_prog(n)%name)//'_advection_on_nrho', &
	Dens%neutralrho_axes(1:3), Time%model_time, 'rho*dzt*advection tendency binned to neutral density',&
	'kg/(sec*m^2)', missing_value=missing_value, range=(/-1.e18,1.e18/))
	id_tracer2_advection(n) = register_diag_field ('ocean_model', trim(T_prog(n)%name)//'_sq_advection', &
	Grd%tracer_axes(1:3), Time%model_time, 'rho*dzt*advection tendency for tracer sqrd',&
	'kg/(m^2*s)*(tracer squared)', missing_value=missing_value, range=(/-1.e18,1.e18/))
	id_sweby_advect(n) = register_diag_field ('ocean_model', trim(T_prog(n)%name)//'_sweby_advect', &
	Grd%tracer_axes(1:3), Time%model_time, 'rho*dzt*sweby advect tendency', &
	'kg/(sec*m^2)', missing_value=missing_value, range=(/-1.e18,1.e18/))
	id_psom_advect(n) = register_diag_field ('ocean_model', trim(T_prog(n)%name)//'_psom_advect', &
	Grd%tracer_axes(1:3), Time%model_time, 'rho*dzt*psom advect tendency', &
	'kg/(sec*m^2)', missing_value=missing_value, range=(/-1.e18,1.e18/))
	id_horz_advect(n) = register_diag_field ('ocean_model', trim(T_prog(n)%name)//'_horz_advect', &
	Grd%tracer_axes(1:3), Time%model_time, 'rho*dzt*horz advect tendency', &
	'kg/(sec*m^2)', missing_value=missing_value, range=(/-1.e18,1.e18/))
	id_vert_advect(n) = register_diag_field ('ocean_model', trim(T_prog(n)%name)//'_vert_advect', &
	Grd%tracer_axes(1:3), Time%model_time, 'rho*dzt*vert advect tendency', &
	'kg/(sec*m^2)', missing_value=missing_value, range=(/-1.e18,1.e18/))
	id_advection_x(n) = register_diag_field ('ocean_model', trim(T_prog(n)%name)//'_x_adv', &
	Grd%tracer_axes(1:3), Time%model_time, 'rho*dzt*x-advective tendency', &
	trim(T_prog(n)%flux_units), missing_value=missing_value, range=(/-1.e18,1.e18/))
	id_advection_y(n) = register_diag_field ('ocean_model', trim(T_prog(n)%name)//'_y_adv', &
	Grd%tracer_axes(1:3), Time%model_time, 'rho*dzt*y-advective tendency', &
	trim(T_prog(n)%flux_units), missing_value=missing_value, range=(/-1.e18,1.e18/))
	id_advection_z(n) = register_diag_field ('ocean_model', trim(T_prog(n)%name)//'_z_adv', &
	Grd%tracer_axes(1:3), Time%model_time, 'rho*dzt*z-advective tendency', &
	trim(T_prog(n)%flux_units), missing_value=missing_value, range=(/-1.e18,1.e18/))
	id_xflux_adv(n) = register_diag_field ('ocean_model', trim(T_prog(n)%name)//'_xflux_adv', &
	Grd%tracer_axes_flux_x(1:3), Time%model_time, 'rho*dzt*dyt*u*tracer', &
	'kg/sec', missing_value=missing_value, range=(/-1.e18,1.e18/))
	id_yflux_adv(n) = register_diag_field ('ocean_model', trim(T_prog(n)%name)//'_yflux_adv', &
	Grd%tracer_axes_flux_y(1:3), Time%model_time, 'rho*dzt*dxt*v*tracer', &
	'kg/sec', missing_value=missing_value, range=(/-1.e18,1.e18/))
	id_zflux_adv(n) = register_diag_field ('ocean_model', trim(T_prog(n)%name)//'_zflux_adv', &
	Grd%tracer_axes_wt(1:3), Time%model_time, 'rho*dxt*dyt*wt*tracer', &
	'kg/sec', missing_value=missing_value, range=(/-1.e18,1.e18/))
	id_xflux_adv_int_z(n) = register_diag_field ('ocean_model', &
	trim(T_prog(n)%name)//'_xflux_adv_int_z', &
	Grd%tracer_axes_flux_x(1:2), Time%model_time, &
	'z-integral of rho*dzt*dyt*u*tracer', &
	'kg/sec', missing_value=missing_value, range=(/-1.e18,1.e18/))
	id_yflux_adv_int_z(n) = register_diag_field ('ocean_model', &
	trim(T_prog(n)%name)//'_yflux_adv_int_z', &
	Grd%tracer_axes_flux_y(1:2), Time%model_time, &
	'z-integral of rho*dzt*dxt*v*tracer', &
	'kg/sec', missing_value=missing_value, range=(/-1.e18,1.e18/))
	id_tracer_adv_diss(n) = register_diag_field ('ocean_model', trim(T_prog(n)%name)//'_adv_diss', &
	Grd%tracer_axes(1:3), Time%model_time, &
	'dissipation of squared '//trim(T_prog(n)%name)//' from advection errors', &
	'[kg/(m^2*sec)]^2', missing_value=missing_value, &
	range=(/-1.e18,1.e18/))
	endif

So for example BGC advective flux terms get the wrong units.

e.g. adic_xflux_adv, eg /g/data/ik11/outputs/access-om2-01/01deg_jra55v140_iaf_cycle4/output882/ocean/oceanbgc-3d-adic_xflux_adv-1-monthly-mean-ym_1991_07.nc has

    long_name:      rho*dzt*dyt*u*tracer
    units:          kg/sec

but the adic tracer /g/data/ik11/outputs/access-om2-01/01deg_jra55v140_iaf_cycle4/output882/ocean/oceanbgc-3d-adic-1-monthly-mean-ym_1991_07.nc has

    long_name:      natural + anthropogenic dissolved inorganic carbon
    units:          mmol/m^3

If the adic_xflux_adv long_name is correct, the adic_xflux_adv units should be (mmol kg)/(m^3 sec), which is roughly mol/sec.

Maybe generic units for _xflux_adv should be given as (tracer units)*kg/sec. Similarly for the other fluxes, tendencies etc.