of the World Meteorological Organisation in order to develop climate data and
metadata rescue activities across the Greater Mediterranean Region.
2.2.2 Models
Measurements provide information on the past and recent climate. To estimate
possible changes of climate parameters, perspective climate models can be applied.
These models can be divided into two main approaches- dynamical and statistical
climate models. Dynamical climate models can be grouped into Global Climate
Models or Global Circulation Models (GCMs), Regional Climate Models (RCMs),
Earth System Models (ESMs), Coupled Atmosphere Ocean Global Climate Models
(AOGCMs), and others. GCMs and AOGCMs are strongly simplified but contain
the most important physical processes describing our climate system. They are
limited to the representation of large scale effects on the global climate due to
changes in greenhouse gas concentration, eruptive volcanoes etc. Their spatial
resolution for the whole globe is from 3
down to 1.2
. RCMs use model output
from GCMs as forcing to simulate the climate at smaller scales for certain regions.
They contain complex model physics and due to their high spatial resolution
from 50 km down to 3 km (0.5
–0.025
) it is possible to reproduce regional
and local effects through the integration of orography and land use. The second
group of climate models follows a statistical approach. Statistical relationships
between large scale processes and local measurements are extended to estimate
future climate and possible changes can be derived very locally. Typical statistical
models are a weather generator, Markov chains, linear regression, or principle
component analysis. Dynamical and statistical models both have their advantages
and disadvantages and the decision of what kind of climate model to use depends
on the application and the specific question to be answered in relation to future
climate change.
For the interpretation of climate change scenarios and a consequent impact
assessment, the consideration of given uncertainties is a fundamental task. Uncertainties arise from imperfect knowledge of physical processes of the climate system
as well as from model limitations due to the numerical approximation of the physical
equations. Many physical processes which operate at scales below the model resolution are integrated into the climate and impact models as assumptions, simplifications
and parameterisations. Furthermore the internal model variability is a reason for
uncertainties in the simulation of climate responses to given forcings (Christensen
et al. 2001). Moreover, uncertainties arise from the internal variability of the climate
system, which is characterised by natural fluctuations in the absence of any radiative
forcing (Hawkins and Sutton 2009). Additionally, a high level of uncertainty of
the observed climate is implied due to measurement errors and sparse station
networks as already described in the previous section. The development of climate
scenarios involves uncertainties due to the estimation of future greenhouse gas and
aerosol emissions, the conversion of emissions to concentrations, the conversion of
20
I. Anders et al.
metadata rescue activities across the Greater Mediterranean Region.
2.2.2 Models
Measurements provide information on the past and recent climate. To estimate
possible changes of climate parameters, perspective climate models can be applied.
These models can be divided into two main approaches- dynamical and statistical
climate models. Dynamical climate models can be grouped into Global Climate
Models or Global Circulation Models (GCMs), Regional Climate Models (RCMs),
Earth System Models (ESMs), Coupled Atmosphere Ocean Global Climate Models
(AOGCMs), and others. GCMs and AOGCMs are strongly simplified but contain
the most important physical processes describing our climate system. They are
limited to the representation of large scale effects on the global climate due to
changes in greenhouse gas concentration, eruptive volcanoes etc. Their spatial
resolution for the whole globe is from 3
down to 1.2
. RCMs use model output
from GCMs as forcing to simulate the climate at smaller scales for certain regions.
They contain complex model physics and due to their high spatial resolution
from 50 km down to 3 km (0.5
–0.025
) it is possible to reproduce regional
and local effects through the integration of orography and land use. The second
group of climate models follows a statistical approach. Statistical relationships
between large scale processes and local measurements are extended to estimate
future climate and possible changes can be derived very locally. Typical statistical
models are a weather generator, Markov chains, linear regression, or principle
component analysis. Dynamical and statistical models both have their advantages
and disadvantages and the decision of what kind of climate model to use depends
on the application and the specific question to be answered in relation to future
climate change.
For the interpretation of climate change scenarios and a consequent impact
assessment, the consideration of given uncertainties is a fundamental task. Uncertainties arise from imperfect knowledge of physical processes of the climate system
as well as from model limitations due to the numerical approximation of the physical
equations. Many physical processes which operate at scales below the model resolution are integrated into the climate and impact models as assumptions, simplifications
and parameterisations. Furthermore the internal model variability is a reason for
uncertainties in the simulation of climate responses to given forcings (Christensen
et al. 2001). Moreover, uncertainties arise from the internal variability of the climate
system, which is characterised by natural fluctuations in the absence of any radiative
forcing (Hawkins and Sutton 2009). Additionally, a high level of uncertainty of
the observed climate is implied due to measurement errors and sparse station
networks as already described in the previous section. The development of climate
scenarios involves uncertainties due to the estimation of future greenhouse gas and
aerosol emissions, the conversion of emissions to concentrations, the conversion of
20
I. Anders et al.