1. #1 DACH, Hamburg 14. September 2007 Models „for“ not „of“ Institute of Coastal Research, GKSS Research Centre Geesthacht, and Meteorological Institute, Hamburg University Hans von Storch

2. #2 DACH, Hamburg 14. September 2007 Hesse’s concept of models Reality and a model have attributes, some of which are consistent and others are contradicting. Other attributes are unknown whether reality and model share them. The consistent attributes are positive analogs. The contradicting attributes are negative analogs. The “unknown” attributes are neutral analogs. Hesse, M.B., 1970: Models and analogies in science. University of Notre Dame Press, Notre Dame 184 pp. Conceptual aspects of modelling

3. #3 DACH, Hamburg 14. September 2007 Validating the model means to determine the positive and negative analogs. Applying the model means to assume that specific neutral analogs are actually positive ones. The constructive part of a model is in its neutral analogs.

4. #4 DACH, Hamburg 14. September 2007 Positive analog Neutral analog Application

5. #5 DACH, Hamburg 14. September 2007 Models are smaller than reality (finite number of processes, reduced size of phase space) simpler than reality (description of processes is idealized) closed, whereas reality is open (infinite number of external, unpredictable forcing factors is reduced to a few specified factors)

6. #6 DACH, Hamburg 14. September 2007

7. #7 DACH, Hamburg 14. September 2007

8. #8 DACH, Hamburg 14. September 2007

9. #9 DACH, Hamburg 14. September 2007 Only part of contributing spatial and temporal scales are selected. Parameter range limited Models represent only part of reality; Subjective choice of the researcher; Certain processes are disregarded.

10. #10 DACH, Hamburg 14. September 2007 Models can not be verified because reality is open. Coincidence of modelled and observed state may happen because of model´s skill or because of fortuitous (unknown) external influences, not accounted for by the model.

11. #11 DACH, Hamburg 14. September 2007 Trivially: all models are “false” (= have negative analogs) Some are really garbage, but many are useful (= have sufficiently many positive analogs and relevant neutral analogs).

12. #12 DACH, Hamburg 14. September 2007 Purpose of models # reduction of complex systems  “understanding” # surrogate reality  realism

13. #13 DACH, Hamburg 14. September 2007 Models for reduction of complex systems identification of significant, small subsystems and key processes often derived through scale analysis (Taylor expansion with some characteristic numbers) often derived semi–empirically characteristics: simplicity idealisation conceptualisation fundamental science approach

14. #14 DACH, Hamburg 14. September 2007 Models for reduction of complex systems good for: constitution of “understanding”, i.e. theory construction of hypotheses Validation: reproduces the gross features of key indicators of a phenomenon and the key processes supposedly relevant for the conceptualizing of the phenomenon; all other processes are disregarded.

15. #15 DACH, Hamburg 14. September 2007 Idealized energy balance, with constant or dynamical albedo β Constant or Randomized transmissivity α

16. #16 DACH, Hamburg 14. September 2007 evolution with slightly randomized transmissivity evolution from different initial values with constant transmissivity and temperature dependent albedo Integration of a zero–dimensional energy balance model no noise with noise

17. #17 DACH, Hamburg 14. September 2007 Models as surrogate reality dynamical, process-based models, characteristics: complexity quasi-realistic mathematical/mechanistic engineering approach

18. #18 DACH, Hamburg 14. September 2007 Dynamical processes in the atmosphere Dynamical processes in a global atmospheric general circulation model

19. #19 DACH, Hamburg 14. September 2007 These models are built to describe the dynamics for certain time and spatial scales and for a certain range of parameters. Validation means that positive analogs prevail for key processes (evolutions, statistics of key variables) Which? … depends in the purpose of the model

20. #20 DACH, Hamburg 14. September 2007 The model can be validated only for that part of the “phase space”, which is sufficiently covered by observations.

21. #21 DACH, Hamburg 14. September 2007 Applying the model outside the admissible domain means to exploit a neutral analog. e.g., climate change scenarios.

22. #22 DACH, Hamburg 14. September 2007 forecast of detailed development (e.g. weather forecast) – neutral analog: future development dynamically consistent interpretation and extrapolation of observations in space and time (“data assimilation”) - neutral analog: space-time correlations reconstruction of global past states and construction of scenarios - neutral analog: sensitivity to external forcings reconstruction of regional past states - neutral analog: dynamical downscaling link process sensitivity analysis – neutral analog: embedding of process in dynamics experimentation tool (test of hypotheses) – neutral analog: all processes significant to the hypothesis are taken realistically into account. Purposes

23. #23 DACH, Hamburg 14. September 2007 Concept of Dynamical Downscaling RCM Physiographic detail 3-d vector of state Known large scale state projection of full state on large-scale scale Large-scale (spectral) nudging

24. #24 DACH, Hamburg 14. September 2007 Positive values show added value of the regional model. 95% significant deviations are marked by a *. Pattern correlations (%) PCC DWD and NCEP PCC improvement/ deterioration RCM Nudge PCC improvement/ deterioration RCM Standard

25. #25 DACH, Hamburg 14. September 2007 Roeckner & Lohmann, 1993 No cirrus Effect of black cirrus detailed parameterization Latitude-height distribution of temperature (deg C) Difference “black cirrus” - detailed parameterization Difference “no cirrus” - detailed parameterization

26. #26 DACH, Hamburg 14. September 2007 Testing the of multimodality of large scale atmospheric dynamics Berner and Branstator, pers. comm

27. #27 DACH, Hamburg 14. September 2007 Testing the MBH “hockeystick method” Simulating the process of “reconstructing” historical climate variations using the data from the 1000 year historical ECHO-G simulation. Done by constructing “pseudo-proxies”. Short-term ( 100 years) severely underestimated. MBH method methodically flawed.

28. #28 DACH, Hamburg 14. September 2007 Conclusions „Models“ can be very different species The different species have different functional properties „Models“ can hardly be verified For further reading, refer to: von Storch, H., S. Güss und M. Heimann, 1999: Das Klimasystem und seine Modellierung. Eine Einführung. Springer Verlag ISBN 3-540-65830-0, 255 pp von Storch, H., and G. Flöser (Eds.), 2001: Models in Environmental Research. Proceedings of the Second GKSS School on Environmental Research, Springer Verlag ISBN 3-540-67862, 254 pp. Müller, P., and H. von Storch, 2004: Computer Modelling in Atmospheric and Oceanic Sciences - Building Knowledge. Springer Verlag Berlin - Heidelberg - New York, 304pp, ISN 1437-028X