FRANCESCHETTI Maël

PhD student
Team : SMA
Arrival date : 10/01/2021
    Sorbonne Université - LIP6
    Boîte courrier 169
    Couloir 25-26, Étage 4, Bureau 403
    4 place Jussieu
    75252 PARIS CEDEX 05
    FRANCE

Tel: +33 1 44 27 87 94, Mael.Franceschetti (at) nulllip6.fr
https://lip6.fr/Mael.Franceschetti

Supervision : Jean-Daniel KANT

TerraNeon: Simulate the impact of individual ecological practices on the climate

The Climate Emergency is no longer to be demonstrated. The year 2020 was the hottest year ever recorded on Earth. The IPCC1 reports indicate without ambiguity the imperative need to halve our greenhouse gas (GHG) emissions by 2030, and by 4 by 2050, if we want to avoid future climatic disasters, which will affect every region of the planet, including France. The TerraNeon project is an ambitious, medium-term, highly multidisciplinary project, aiming to build the basis of a simulator of the impact of human activities on the climate and in particular global warming. There are many works allowing to simulate and predict the evolution of the climate based on physical or statistical models (e.g. IPCC), and most of them include simulation variables of human activities. But, to our knowledge, these human behaviors are not modeled individually but aggregated in the form of exogenous impact parameters, the set of human individuals being thus aggregated in the form of a typical behavior, in the manner of the representative agent model as often practiced in economics. If these numerical climate and environmental models can provide precise information on the possible trajectories of the climate and the environment, they do not sufficiently detail the variability of human, economic and social behaviors that can affect them. However, the climate emergency requires us to be able to determine as scientifically as possible which individual and collective behaviors would be the most harmful or virtuous. Having this kind of information would allow us to raise awareness and help each actor in society (decision-maker, individual, company, institution, etc.). It would contribute to the identification, design and implementation of the best actions and policies likely to change the trajectory, to avoid the predicted disasters and improve our future. For this, we propose a hybrid and innovative approach, in order to couple the best numerical models of climate and environment with multi-agent models of human behavior (individual, social and economic). The recent progress of multi-agent-based modeling and simulation techniques, both for economic and social simulation, make such an approach possible today. Sorting your waste, eating less meat, not flying anymore, insulating a building, ..., this is a non-exhaustive list of individual actions that are often cited to fight against global warming. Unfortunately it is not possible today to automatically identify the rate of adherence necessary to make their impact significant, nor to estimate their economic cost or social acceptability. In this project, we aim to build an innovative decision support tool to : - study the emerging threshold effects for these different IEP practices: what would be the minimum proportion of people required for an action to start having a significant effect on global warming (and in particular on greenhouse gas (GHG) emissions)? - to understand the impact of social interactions on adoption: diffusion of innovations of innovations, dynamics of opinions... - evaluate the effects of incentive policies (communication, nudging, subsidies...) to increase adoption. - to automatically find out which policies will reduce the GHG emissions of individuals in the most efficient way. For this purpose, we will couple the TerraNeon simulator to our optimization tool OptiPol. Simulating the impact of human interactions and showing that a measure is virtuous (or optimizing it so that it becomes virtuous) will help convince actors and decision-makers to apply it. Indeed, our hypothesis is that individuals find it difficult to adopt ecological practices for three main reasons 1. their estimated, certain, individual and immediate cost, compared to a collective, hoped-for and distant benefit collective, hoped-for and distant benefit. 2. they are not convinced that an individual effort will have an impact on global warming. In fact, following the example of T. Schelling's famous work on urban segregation, only a simulation will be able to show the multiplier effects of each actor's actions (individual, company, institution) on the climate. Simulation will allow us to show decision-makers, consumers and citizens the effects of their actions on the climate, and to explain them in order to better convince them of how we can collectively limit our impact on the climate. Multi-agent simulations have the enormous advantage of being "white boxes" that can be fully explained, unlike other competing approaches (neural networks, etc.)