Enhancing ClimRisk: using machine learning to better predict systemic social risks from climate change 

The challenge

The consequences of climate change are evident in rising global temperatures, more frequent and severe weather events, shifts in rainfall patterns, and rising sea levels. These changes pose significant threats to ecosystems, biodiversity, and society.

As our understanding of climate change has developed, we have identified and defined climate risks. Climate risks are the impacts of manifest changes to the frequency and intensity of extreme events – such as heat waves, drought, tropical cyclones and more – on our socio-economic, health and geo-political structures.

In this context, resource scarcity amplified by climate change is likely to play a greater role in making conflict worse across the Asia Pacific region, and beyond.  At the moment, analyses of climate-induced conflicts are largely conducted after the fact, through qualitative case study description.

Our challenge is how best to integrate socio-economic data with our physical climate models. We need this type of combined multi-factor approach to predict and mitigate the impact of climate risks on our social and geo-political systems into the future.

Our response

CSIRO’s Responsible Innovation Future Science Platform and Environment Business Unit are collaborating to explore these challenges and build on the team’s existing state-of-the art research on predicting the impacts of physical climate risks. Beyond the immediate physical and socio-economic consequences of our changing climate, we also need to account for systemic, cascading risks such as geo-political unrest.

The research team have already established a software platform, ClimRisk, which draws on the combined power of machine learning techniques coupled with climate prediction. This project now seeks to extend the capability of that software platform.  We’ll do this by applying the latest methods from machine learning, dynamical systems and statistical physics to real-world climate and socio-economic data sets.

The team will identify the causal relationships along the chain of influence from carbon emissions to climate, to agricultural production, commodity prices, food affordability, and ultimately the potential for social unrest and conflict. A core part of our approach involves quantifying the impact of the climate on historical conflicts and social unrest, to determine what elements might be predictable and over what future time horizon to develop a new predictive capability.

Impact

The goal of this research is to produce robust estimates of climate-induced risks to the social and economic structures that underpin our nation’s security. It also aims to enhance our ability to successfully navigate oncoming impacts.

In this project, we focus on developing new quantitative tools and techniques for identifying regions where there is elevated risk of social unrest or conflict in the near term due to climate induced resource scarcity.

This will enable better informed decision-making and responsiveness, for example by proactively sending aid to flagged regions, or relocating people to lower risk areas early enough to prevent loss of life.

The potential to build a better warning system for climate-induced social conflict using these methods also holds immense potential for a range of other applications. For example, predicting the likelihood and impact of climate-induced health crises, and other cascading system-level effects.

Team

Terry O’Kane (Project Lead), Vassili Kitsios, David Newth and Dan Li.

Related research

Causal inference in complex multiscale systems: https://research.csiro.au/ai4m/causal-inference-in-complex-multiscale-systems/

Links

Seminar (27 July 2022) by Dr Vassili Kitsios to the University of Melbourne’s School of Geography Earth and Atmospheric Sciences on the quantification of climate risk in health care and commodity markets