Impact of atmospheric models' horizontal resolution on water budgets in HighResMip

VANNIERE, BENOIT; VIDALE, PIER LUIGI; ROBERTS, MALCOLM; DEMORY, MARIA-ESTELLE; HODGES, KEVIN; MÜLLER, OMAR V.

San Francisco

Conferencia; AGU Fall Meeting 2019; 2019

AGU

Processes governing the global hydrological cycle cover all scales from the synoptic to the microphysical scales. At one end of the spectrum, synoptic systems are fully resolved by current GCMs, whereas at the other end microphysics is fully parameterised. When the resolution of atmospheric models is increased from 1 degree, as typically used by CGCMs, to 0.25 degree, as made possible by advances in computer performance and recent model development, new processes emerge in the so-called mesoscale range, which were not accounted for when developing parameterisations (e.g. sharper atmospheric fronts, richer eddies).Many questions arise with their emergence, but in this paper we ask: to which extent do these emerging processes change the global water budget? The European PRIMAVERA project has offered an unprecedented opportunity to answer this question by gathering an ensemble of eight European CGCMs and following the HighResMip protocol.We find that the global hydrological cycle intensifies at high resolution, with more surface evaporation occurring in midlatitudes. The fate of this extra moisture depends on model formulation. In spectral models, precipitation increases over the ocean, but in grid-point models precipitation increases, mostly over high orography. As a consequence, the latter models simulate a dramatic increase of moisture advection to land with increasing resolution. The assessment based on PRIMAVERA models has led us to question recent global estimates of the hydrological cycle based on observations, which could underestimate moisture advection to land, orographic precipitation and river runoff altogether; our evidence is based on a careful comparison of observed and simulated river runoff at the catchment scale.A water budget has also been conducted at the scale of Tropical Cyclones (TCs). Interestingly, we find that the total amount of precipitation associated with TCs is largely constrained by the large-scale moisture flux convergence, which is relatively well captured by GCMs, even at low resolution. However, several aspects of simulation are still unsatisfactory at low resolution: the lack of TC genesis compromises the simulation of the TC contribution to global precipitation, while the too spread-out structure of TCs degrades realism in the simulation of extreme precipitation.