AMM7 25 hourly mean bug

[hadjt]

There appears to be a bug in the 25hour mean calculation in AMM7 (but not AMM15) in the operational system, and the Reanalysis.

Calculating the 25 hour means from the instantaneous hourly SSH gives negligble difference compared to the output 25 hour mean in AMM15 (the difference of the order of +-1e-8m, with no spatial pattern.

Calculating the 25 hour means from the instantaneous hourly SSH gives small difference compared to the output 25 hour mean in AMM7 (the difference of the order of +-4cm, with a clear spatial pattern of the tide.

The DIA/dia25hm.F90 code is identical (in our trial version anyway). It appears to be the difference intialisation field, from the Intial Conditions, and it looks like it one time step later in AMM7 (error in appears lowest 1 times step (300sec) after the 0th hour). I suspect the sshb = sshn, sshn = ssha vs where the dia25hm means are initialised is different between AMM7 and AMM15, maybe a name list setting??

e.g. from the reanalysis, for metoffice_foam1_amm7_NWS_SSH_dm20250516.nc. Upper left and right are the diagnosed 25hour mean, and the post processed 25 hour mean, bottom left is the difference, and bottom right is the ratio.