Specific yield (sy) not present in composite sensitivities

[mgozdor2]

Hi all,

I’m working through MF6-adjoint and running a transient simulation (~30 days) with a head-based performance measure.

I’m computing normalized sensitivities using the composite outputs in the adjoint HDF5 file.

Model setup
MODFLOW 6 transient simulation (30 days)
STO package active
Top 6 layers are convertible (icelltype = 1)
Specific yield (sy) is spatially uniform (0.3 in upper layers)
Specific storage (ss) also defined
What I observe

In the adjoint HDF5 file (adjoint_solution_phi_heads.hdf5):

I see composite sensitivities for:

• k11, k33
• rch6_recharge
• ss
• wel6_q
• etc.

However, there is no composite/sy dataset.
Searching the HDF5 file shows no keys containing sy at all.
STO is present and gwf.sto.sy.array has expected values.
Convertible cells exist in upper layers (millions of cells).
Head change over the simulation is small (~0.001 ft or less in upper layers).

So, is the absence of sy in the composite sensitivities expected behavior?

Specifically:

Is sy currently supported in MF6-adjoint sensitivity output?
Does MF6-adjoint internally lump storage sensitivity into ss?
Or is sy omitted unless certain conditions are met (e.g., larger head changes, water-table movement)?

Basically I want to understand whether:

• sy sensitivities are being computed but not exported

• sy is not currently supported in the adjoint implementation

• Or this is expected behavior for my simulation setup

Any clarification would be greatly appreciated.