Max Planck Institute for Meteorology University of Hamburg Model & Data German High Performance Computing Centre for Climate- and Earth System Research Zeit Foundation Max Planck Society



S_23 Philosophy of Science – part III: The role of quasi-realistic models in building knowledge
Organiser Hans von Storch, Jin-Song von Storch, Corinna Schrum, Frank Lunkeit and Jochem Marotzke
Credits 2
Mode Weekly seminar
Location Geomatikum
Day(s) Wednesdays
Time 16:15 - 17:45
Content Conceptual or maximum reduced models are the ultimate goal of most
scientific endeavours. In fact, such models describe the essentials of the dynamics representative for certain phenomena. They are of minimum complexity, and form the theories of our field. In a sense, the term “to understand a phenomenon” means to have a conceptual model available to explain consistently the characteristic features of a phenomenon.

Such models can be constructed in different ways. In most cases, a mix of
these approaches is used.

• The basic hydro- and thermodynamic equations may be understood as specifying the change of the state variables as a response to a series of „processes“ given by mathematical terms.
These equations become simpler and simpler if more and more terms of the original equations are deleted or approximated by much simpler terms. Here, the art is to find a strategy for identifying the relevant and less relevant processes. Often scalearguments are used to this end, i.e., observational evidence (the scale) is used to assess the relative importance of the many processes. An example are the shallow water equations, or Lorenz’ low order model
• Another approach is to first spatially integrate and secondly simplify the equations. In this case, empirical constants are
introduced into the models, and empirical knowledge is used to determine appropriate values of these constants. Examples are energy balance models or box models.
• Also, more ad-hoc approaches are used. Based on an intuitive understanding, a number of state variables and processes are hypothesized; again, needed constants are derived by fitting the model to observations. Examples are the stochastic climate model, or the delayed action oscillator model of ENSO.
• Statistical analysis is used to identify dominant processes in the simulation with quasi-realistic models; more or less systematic
approaches are used to link these processes.

In most cases, scientific progress is made when at least two of the following three methods are combined: theoretical modelling, quasirealistic modelling and statistical analysis. In this seminar, we will discuss cases of how theoretical knowledge is combined with the other two approaches. Discussed are examples of theoretically guided analysis of the output of quasi-realistic (maximum complexity) models and of complex data sets; of the falsification of theoretical models by performing experiments with maximum complexity models; of the derivation of
hypotheses of dominant processes in a phenomenon.
This seminar is the third in a series of four seminars. They all deal with the problem of how to build scientific progress (theory of science).

The first seminar deals with statistical analysis; the second with maximum complexity models, the third (this one) with minimum complexity models, and the forth one with field campaigns.
   
Comments  

05.03.2012 08:00

Call 2012-II closed on 29 February

We will re-open our Call for Applications on 1 July 2012.


Cat: General

Have a look at our Alumni News.

Have a look at our Alumni Portal.