Geoscience Reference
In-Depth Information
local evaporation the dominant source of model precipitation for most of the year, instead
of moisture convergence from large-scale circulation as suggested by reanalyses. This may
be related to model resolution, as discussed in the next section.
2.2 The Role of Model Resolution
It is widely believed that scale interactions (between short/small and long/large scales in
time and space, and vice versa) are an important feature of the climate system, particularly
in the tropics (e.g., Slingo et al.
2003
). Consequently, a change in the resolution of a
climate model can be expected to affect the simulated climate, not only at the scale of the
grid point/time-step but also at much larger scales. Examples where such impacts have
been documented include the simulation of ocean eddies, and associated improvements in
the mean ocean state, in the Hadley Centre Coupled Eddy-permitting model (HadCEM;
Roberts et al.
2004
), improvements in the simulated tropical climate due to the repre-
sentation of small-scale atmosphere-ocean interactions in the High-resolution Global
Environment Model (HiGEM; Shaffrey et al.
2009
), and improvements in mid-latitude
variability through reduced North Atlantic SST biases (Scaife et al.
2011
).
These and many other studies have resulted in several modelling centres actively
pursuing increased horizontal and vertical resolution in various components of both
regional and, increasingly, global models for a range of timescales. The results of some of
these advances are mentioned in the sub-sections of Sect.
2.3
, while, in the next sub-
sections, we focus specifically on the relationship between model resolution and global
precipitation, and the issue of resolved versus parametrised processes.
2.2.1 Influence of Resolution on Global Precipitation
In order to understand the systematic impact of model resolution in the simulation of the
hydrological cycle, it is useful to use hierarchy of models with the same physical configu-
ration. Demory et al. (
2012
) used a series of simulations of the MetUM climate configuration
HadGEM1 (Martin et al.
2006
) and a prototype configuration of HadGEM3 (Hewitt et al.
2011
), with a wide range of horizontal resolutions, to investigate the impacts of resolution on
the mean simulated model precipitation at large spatial scales. Demory et al. (
2012
) found that
there is a redistribution of precipitation from ocean to land with resolution and that the ratio of
EtoP over land decreases with resolution (Fig.
2
, black dots and circles), bringing the ratio
closer to the range of observed values estimated by Trenberth et al. (
2007
,
2011
). This
suggests that higher resolution models decrease moisture recycling over land.
These changes are associated with an increase in the transport of moisture from ocean to
land at higher resolution (red squares on Fig.
2
). By decomposing the moisture conver-
gence into mean and transient (which is related to eddies) terms, it has been found that the
increased moisture convergence mainly comes from the mean flow, but that the transient
term, associated with storminess, becomes more important with increased resolution
(M.-E. Demory, personal communication, May 2012). Demory et al. (
2012
) conclude that a
resolution of at least 50 km is necessary to simulate the mean global hydrological cycle.
2.2.2 Resolved Versus Parametrised Processes in Models
With the increase in computing power in recent years, the influence of resolving hydro-
logical processes, compared with using physical parametrisations, on the distribution and
variability of precipitation in global models has become a key area of interest among
