RAMDAM methodological axis

RAMDAM  : Navigating the theory-data-model continuum

MODERATOR : Philippe CHOLER

Remember that all models are wrong ; the practical question is how wrong do they have to be to not be useful

Box, G. E. P. and N. R. Draper (1987). Empirical Model-Building and Response Surfaces

 

RAMDAM is a cross-disciplinary action addressing conceptual and methodological challenges in ecological modelling. Although the question of modelling is central in science, there is still vivid debate on what we expect from models.
As for other disciplines, ecology and evolution have now entered the information age and "big data" era1. The rise of massive data (environmental DNA, megaphylogenies, remote sensing, advanced sensors and sensor networks, distributed experiments etc) is accompanied by unprecedented developments in data analysis and modelling.

 

The way we are intertwining theory, data and models is changing. For example, the efficacy of data-driven approaches may question the usefulness of theoretical and physically-based models.

 

RAMDAM will address the main following questions :

 

– what are the opportunities offered by big data and data-intensive approaches for a better understanding of biodiversity dynamics and ecosystem functioning ?

– to what extent the structure of existing models (s.l.) can accommodate with this wealth of new information ?

– how can we leverage our time and effort to advanced field observations and modelling to better inform on ongoing ecological changes ?

 

These questions are discussed through three kind of brainstorming sessions that connect people of different backgrounds.

 

1. Bimonthly internal meetings dedicated to the dissemination of knowledge and know-hows on data modelling and concepts. This relies on the expertise present in the lab and capitalizes on the existing workshops of the R community users. The sessions will be introduced by a short presentation on an emerging method (e.g. Hierarchical models - Structural Equation Models – Approximate Bayesian Computation etc) or a novel source of data (high-resolution imagery, gridded data set of climate etc..) and its specific use for research questions addressed in the lab.

2. Quarterly working sessions animated by a guest speaker to introduce data analyses and/or modelling approaches that are not yet commonly used in the lab. Whenever possible, this is organized in partnership with the other labs of the FREE or the OSUG federation. We are anticipating strong links with the activities of the Multidisciplinary Institute of Artificial Intelligency (MIAI@UGA), and especially its chairs dedicated to environment.

3. Half yearly open forums for discussion and debate on various epistemic issues raised in climate and biodiversity sciences and their links with major societal problems : status of models, verification and validation, confidence and uncertainty, knowledge and decision etc. These discussions will be introduced by the critical reading of a paper or a book in a "journal club" format.

 

Qhat are the on-going projects ?

Projects
248