Details zur Publikation |
| Kategorie | Textpublikation |
| Referenztyp | Zeitschriften |
| DOI | 10.1016/j.ecolmodel.2015.07.018 |
| Titel (primär) | Robustness analysis: Deconstructing computational models for ecological theory and applications |
| Autor | Grimm, V.; Berger, U. |
| Quelle | Ecological Modelling |
| Erscheinungsjahr | 2016 |
| Department | OESA; iDiv |
| Band/Volume | 326 |
| Seite von | 162 |
| Seite bis | 167 |
| Sprache | englisch |
| Keywords | Sensitivity analysis; Ecological theory; Computational modelling; Robustness; Model analysis; Understanding |
| UFZ Querschnittsthemen | RU5; |
| Abstract | The design of computational models is path-dependent: the choices made in each step during model development constrain the choices that are available in the subsequent steps. The actual path of model development can be extremely different, even for the same system, because the path depends on the question addressed, the availability of data, and the consideration of specific expert knowledge, in addition to the experience, background, and modelling preferences of the modellers. Thus, insights from different models are practically impossible to integrate, which hinders the development of general theory. We therefore suggest augmenting the current culture of communicating models as working just fine with a culture of presenting analyses in which we try to break models, i.e., model mechanisms explaining certain observations break down. We refer to the systematic attempts to break a model as “robustness analysis” (RA). RA is the systematic deconstruction of a model by forcefully changing the model's parameters, structure, and representation of processes. We discuss the nature and elements of RA and provide brief examples. RA cannot be completely formalized into specific techniques and instead corresponds to detective work that is driven by general questions and specific hypotheses, with strong attention focused on unusual behaviours. Both individual modellers and ecological modelling in general will benefit from RA because RA helps with understanding models and identifying “robust theories”, which are general principles that are independent of the idiosyncrasies of specific models. Integrating the results of RAs from different models to address certain systems or questions will then provide a comprehensive overview of when certain mechanisms control system behaviour and when and why this control ceases. This approach can provide insights into the mechanisms that lead to regime shifts in actual ecological systems. |
| dauerhafte UFZ-Verlinkung | https://www.ufz.de/index.php?en=20939&ufzPublicationIdentifier=16775 |
| Grimm, V., Berger, U. (2016): Robustness analysis: Deconstructing computational models for ecological theory and applications Ecol. Model. 326 , 162 - 167 10.1016/j.ecolmodel.2015.07.018 |
|