Context

Misleading output can be generated from dia_hth with some combinations of output diagnostics

Analysis

The code section which initialises the temporary arrays is contained within a:

       IF( iom_use( 'mlddzt' ) .OR. iom_use( 'mldr0_3' ) .OR. iom_use( 'mldr0_1' ) ) THEN

statement but it is possible to request other diagnostics whilst omitting any of these three. In this case the arrays are not initialised and results can contain random values. See, for example: ​https://nemo-ocean.discourse.group/t/4-0-x-bug-in-variable-output-mldr10-3/230

...

Recommendation

Extend the IF statement so that the initialisation is triggered if any of the involved variables are active.

Context

Conservative update with AGRIF and non linear free surface

Analysis

During the thickness weighted update stage, wrong scale factors at before time step are used. This leads to a small constancy preservation issue.

Recommendation

Use "before" scale factors prior update.

Context

correction specific to the write of averaged Kz in diawri.F90, such as for offline BIO.

Analysis

Kz must be log-transformed before being averaged. The logarithm to the base e (LOG) was used instead of the logarithm to the base 10 (LOG10)

Recommendation

change LOG by LOG10 in diawri.F90:

IF( iom_use('logavt') ) CALL iom_put( "logavt", LOG10( MAX( 1.e-20_wp, avt(:,:,:) ) ) ) IF( iom_use('logavs') ) CALL iom_put( "logavs", LOG10( MAX( 1.e-20_wp, avs(:,:,:) ) ) )