Numerical Simulations of Short-Term Non-tidal Ocean Mass Anomalies 123
significant changes in precipitation trends occur within ERA-40, which are con-
nected with the advent of new observing systems rather than with real changes
of mass transports in the hydrological cycle. Moreover, due to significant over-
estimations of precipitation over tropical oceans within the atmospheric model,
precipitation exceeds evaporation over oceans as well as over land. As indicated
by Andersson et al. (2005), improvements of clouds and rain assimilation in the
ECMWF forecasting s ystem will take place in the near future, which will probably
result in more realistic net freshwater fluxes from ECMWF operational data. Despite
these errors, data from weather prediction models currently provide the only oppor-
tunity to examine the transient impact of freshwater fluxes on ocean circulation in
an operational model set-up.
To ensure consistent water mass fluxes among the major sub-systems of the
Earth, river discharges used in this study are based on the same atmospheric
data, which are also used to force the ocean model OMCT. Geographically dis-
tributed river discharge data have been obtained from numerical simulations with the
Hydrological Discharge Model (HDM; Hagemann and Dümenil, 1998). This model
is part of the coupled atmosphere-ocean global circulation model ECHAM5/MPI-
OM (Latif et al., 2003) and has been used in several studies, e.g., to validate the
hydrological cycle of ERA-40 (Hagemann et al., 2005) and to analyse the impact
of hydrological mass variations on the Earth’s rotation (Walter, 2007). HDM is a
linear cascade model with a constant horizontal resolution of 0.5
◦
in longitude and
latitude and uses a time step of 1 day. The model is capable of simulating lateral dis-
charge in 3 layers: overland flow, river flow and ground water flow. Applied forcing
fields are daily estimates of run-off and drainage, which have been obtained from
2m-temperatures and precipitation provided by ERA-40 and ECMWF’s operational
data sets using a land-surface scheme (Hagemann and Dümenil, 1998). The qual-
ity of simulated river flow has been validated by means of gauge data provided by
the Global Run-off Data Centre (Koblenz, Germany). According to Walter (2007),
annual mean discharges as well as seasonal variations are reproduced realistically
by HDM.
Based on OMCT simulations covering the period 1958–2005, the impact of
time-variable freshwater fluxes on ocean mass variability has been evaluated. Mass
varability due to changes in total ocean mass have been corrected for, since they
will be considered separately in the following section. Generally, the direct effects
of freshwater fluxes on ocean mass transports are small (Fig. 2a), since freshwater
anomalies primarily cause variations in the density structure and therefore in steric
sea-level (see, e.g., Dobslaw, 2007). Thus, there are consequences on ocean bot-
tom pressure only in the rare case of density changes throughout the whole water
column down to the bottom, or indirectly due to changes in the thermohaline cir-
culation. Significant mass variability due to freshwater fluxes is therefore confined
to shallow coastal areas, while mass re-distributions of most parts of the oceans are
not affected, provided that the long-term stability of the thermohaline circulation is
maintained by, e.g., relaxation of the sea surface salinity towards an observational
climatology (Conkright et al., 2002). For this particular analysis, the relaxation time-
scale has been extended to 180 days in order to avoid influences of the relaxation