Deep reinforcement learning in World-Earth system models to discover sustainable management strategies

Abstract

Increasingly complex, non-linear World-Earth system models are used for describing the dynamics of the biophysical Earth system and the socio-economic and socio-cultural World of human societies and their interactions. Identifying pathways towards a sustainable future in these models for informing policy makers and the wider public, e.g. pathways leading to a robust mitigation of dangerous anthropogenic climate change, is a challenging and widely investigated task in the field of climate research and broader Earth system science. This problem is particularly difficult when constraints on avoiding transgressions of planetary boundaries and social foundations need to be taken into account. In this work, we propose to combine recently developed machine learning techniques, namely deep reinforcement learning (DRL), with classical analysis of trajectories in the World-Earth system. Based on the concept of the agent-environment interface, we develop an agent that is generally able to act and learn in variable manageable environment models of the Earth system. We demonstrate the potential of our framework by applying DRL algorithms to two stylized World-Earth system models. Conceptually, we explore thereby the feasibility of finding novel global governance policies leading into a safe and just operating space constrained by certain planetary and socio-economic boundaries. The artificially intelligent agent learns that the timing of a specific mix of taxing carbon emissions and subsidies on renewables is of crucial relevance for finding World-Earth system trajectories that are sustainable on the long term.We propose a framework for using deep reinforcement learning (DRL) as an approach to extend the field of Earth system analysis by a new method. We build our framework upon the agentenvironment interface concept. The agent can apply management options to models of the Earth system as the environment of interest and learn by rewards provided by the environment. We train our agent with a deep Q-neural network extended by current state-of-the-art algorithms. We find that the agent is able to learn novel, previously undiscovered policies that navigate the system into sustainable regions in two exemplary conceptual models of the World-Earth system.

Modeling the effects of environmental and perceptual uncertainty using deterministic reinforcement learning dynamics with partial observability

“…We also showed that partial observability can lead to better collective outcomes in the case of social dilemmas. The question for the preconditions of cooperation and sustainable behavior presents an important area for deeper investigation [79][80][81]. Temporal-difference learning is a widespread principle in neuroscience and psychology [44] and there is indeed evidence that humans use a payoff-based learning rule in social dilemmas [82].…”

Taxonomies for structuring models for World–Earth systems analysis of the Anthropocene: subsystems, their interactions and social–ecological feedback loops

“…Current efforts in World-Earth systems modelling are highly stylised (e.g. Kellie-Smith and Cox, 2011;Garrett, 2015;Jarvis et al, 2015;Heck et al, 2016;Nitzbon et al, 2017;Strnad et al, 2019) or tend to be proof-of-concept prototypes (Beckage et al, 2018;Donges et al, 2020). None currently operate in a processdetailed, well-validated and data-driven mode.…”

Optimization of coupling and global collapse in diffusively coupled socio-ecological resource exploitation networks

“…Through a combination of techniques from social and ecological network models [10,[27][28][29] complex networks have recently been proven as a promising approach to bridge theoretical physics and efforts to understand future trajectories of the Earth system in the Anthropocene [30][31][32], where human social dynamics have become a dominant geological process [33,34]. These so-called World-Earth models [31,35,36] have for example been used to study emergent characteristics of interactions between social networks of resource harvesting agents [27][28][29], impacts of multi-agent social learning and market dynamics on deforestation rates in rain forests [37], or the emergence of sudden regime shifts in socio-ecological systems driven by specific network characteristics [38,39].…”