Progress in Earth System Science: What Does It Take to Make Our Ideas Clear? - Towards an Interdisciplinary Approach in Earth System Science - page 25

Fig. 1 Schematic view of
evidences, data, modelling,
interpretation. The progress is
usually an iterative loop with
hypotheses testing and
developing ideas

Data
(old, new,
preliminary)

Idea Analysis

Model
(preliminary, then
improved)

Results
(preliminary, then critically
improved)

CO 2 levels) and study the direction and magnitude of response (such as an increase
in sea level or ocean acidi cation), which is constrained by physical laws, obser-

vations or other models. Based on the state of our current knowledge, climate
simulations give the most probable outcome under different scenarios.
Important to note, the type of model used depends on the nature of, and the
timescales relevant to, the question asked. On the one extreme are the relatively
simple Conceptual Models that contain a very limited number of processes, but
which clarify the key features of a system (Fig. 2 ). On the other extreme, Com-
prehensive Models require super computers and large research teams for mainte-
nance (such as General Circulation Models, so called GCMs) operating at a high
spatial and temporal resolution (e.g., Peixoto and Oort 1992 ; McGuf e and

Henderson-Sellers 2014). Between these two extremes are the Models of Inter-
mediate Complexity (Claussen et al. 2002 ). The transition from highly complex
dynamical equations to a low-order description of climate is an important step since
it gives more credibility to the approach and its results. Low-order systems can
clarify the main effects in the system, neglecting all the second order processes and
selecting the speci c temporal and spatial scales, or can be used as a data-inter-

pretation tool using statistical-conceptual models analyzing the complex observa-
tional system (e.g., Stommel 1961 ; Hasselmann 1976 ; Lemke 1997 ; Rooth 1982 ;
Lohmann and Schneider 1999 ; Stocker and Johnsen 2003 ; Laepple et al. 2011 ).
Our concepts become clear if we describe the phenomena on different levels of
model hierarchies as sketched in Fig. 2 . In his topic, Saltzman ( 2002 ) formulated a
dynamical system approach in order to differentiate between fast-response and
slow-response variables for long-term climate variability and change. One
straightforward approach is coarse graining where the underlying dynamics are
projected onto the macroscopic dynamics, which are used in statistical physics
theory of non-equilibrium statistical mechanics (e.g., Zwanzig 1960 ; Mori et al.
1974 ). A similar concept in climate sciences had already been proposed by Has-
selmann ( 1976 ), in which the climate system is described by climate variables (in
statistical physics they are called “ macroscopic ” ) and weather ( “ microscopic ” )