Geoscience Reference
In-Depth Information
-
Pre-processing
: this phase includes carrying out quality control on the
observed data that is exchanged globally through the Global Telecommunications
System (GTS) implemented by the World Meteorological Organization (WMO).
This phase also involves transforming the data that are to be used into the format
required by the data assimilation scheme.
-
Objective analysis
: this is the phase enabling definition of an initial 3D
representation of the climate system over the considered area (global or limited
area), from data that are collected, and also from a recent forecast which is used as a
first guess.
-
Data assimilation
: consists of establishing a series of analyses consistent with
observations. The amount of observations taken into account is linked to the so-
called cut-off time, i.e. the time allowed after the synoptic hour considered for
collecting subsequent data. Data assimilation is meant to collect as much data as
possible needed for describing, as well as possible, the initial state of the atmosphere
and of other considered interacting components of the climate system, from which
the forecast will be made. Most commonly used methods include optimal
interpolation and variational assimilation. A special version of the latter has been
developed to deal with non-synchronous observations (in other words 4D-VAR).
-
Physiographical data
: orography, land and sea areas, aerodynamic roughness,
type of land and vegetation cover; surface albedo; etc, are geographical data that are
generally maintained constant from the initial state to the final lead time for weather
prediction purposes.
-
The initialization
is used to make the structure of observed meteorological
fields consistent with the models that are used. It is also used to filter the
development of oscillations generated by gravity waves, while retaining the most
important meteorological structures. Commonly used techniques include digital
filtering and normal-mode initialization.
-
The weather forecast model itself
: the evolution of atmospheric fluid is
submitted to external forces (gravity, Coriolis force due to the Earth's rotation, and
friction). It is also affected by internal forces, such as pressure differences and
buoyancy. Both set of forces obey the laws of physics, as well as the laws of fluid
dynamics, including conservation of mass and energy, and momentum.
In theory, the laws of temporal evolution of the different variables describing the
structure of the atmospheric circulation make it possible to predict how the fluid will
evolve, provided its initial state is known.
Due to uncertainties at all levels, ensembles of forecast are often used instead of
a single forecast. For example, as the development of severe weather is frequently
highly non-linear and therefore sensitive to forecast errors, this is an appropriate
application of ensembles.
The simulations start with an initial state that is slightly perturbed in different
ways, the produced ensemble is then subject to probabilistic analyses. The
atmosphere is divided into a large number of elementary boxes (Figure 4.4) which
cover either the entire world (global models), or a smaller zone (limited area model).
