Grell 3D scheme

Grell 3D (G3) is the WRF implementation that carries the Grell-Devenyi ensemble closure approach forward, with one substantive addition: subsidence spreading. Compensating subsidence, which conventional mass-flux schemes confine to the column containing the updraft, can be distributed across neighbouring grid columns.

That option is the reason the scheme exists as a separate choice, and it addresses a real inconsistency. A mass-flux scheme requires that the updraft be compensated by subsidence somewhere. Putting all of it in the same column is a modelling convenience justified by the cell being large compared with the convection. As resolution increases that stops being true — real compensating subsidence occurs over a region substantially wider than the updraft, spanning several grid cells at fine spacing. Confining it to one column concentrates warming and drying artificially and can produce grid-point storms, where convection locks onto a single cell and intensifies unphysically.

Spreading subsidence across neighbours relieves this. It is a partial answer to the same grey-zone problem Grell-Freitas addresses through scale-aware mass flux, arrived at from a different direction — one modifies how much convection happens, the other where its compensation goes.

Choosing among the Grell family

The three are best understood as a sequence rather than alternatives:

• Grell-Devenyi established the ensemble-of-closures idea.
• Grell 3D is the WRF-integrated version with subsidence spreading, and is the practical default of the three for conventional resolutions.
• Grell-Freitas adds scale awareness and is the one to use across nests spanning very different grid spacings.

As with any convective closure, the choice should be tested against precipitation timing and intensity in the target region rather than inherited from a previous study. Convective schemes are calibrated, implicitly, against the regimes their developers worked in, and a scheme that performs well over the continental United States has no obligation to do so over a tropical archipelago. In the DARWIN experiment suite the convection scheme was treated as a variable to be tested rather than a setting to be chosen once.

See also: cumulus convection schemes, Grell-Devenyi, Grell-Freitas, and Kain-Fritsch for the single-closure family.