Climate change, social inequality & psychosocial wellbeing with emerging digital data - a multidisciplinary network between UK and South Korea

Lead Research Organisation: University of Leeds
Department Name: Leeds University Business School (LUBS)

Abstract

Climate change impacts are not only leading to physical consequences in terms of direct threats to life, infrastructure, and the natural environment, but also have profound social implications and threaten to widen existing inequalities. Climate change-related inequalities in psychosocial wellbeing (e.g., exposure to the severe distress triggered by extreme weather events, eco-anxiety of 'no future'), are relatively difficult to analyse directly because they are typically 'inner' (e.g., subjective experiences and feelings) at the individual level. Recent developments in social data science and machine learning techniques, mean the disproportionate effect of climate change on different vulnerable groups can be measured and tracked using emerging sources of digital data such as street-view images, user-generated content, crowdsourced data, mobile phones, and social media. These new data sources contain patterns of various human behaviours in vulnerable people, which can be utilised to infer individual differences in psychosocial wellbeing. This data makes it feasible to understand how and to what extent climate change exacerbates social inequalities by affecting psychosocial wellbeing, and what protective factors and support requirements are needed. It also provides a new lens to study and address climate change and psychosocial wellbeing inequality within a uniform data environment and technique framework. Developing future collaborative research programmes based on this has the potential to assist in the achievement of UN's SDG 11(where inequality reduction is a core) and 13 together, which also aligns with UK and South Korea government priorities.

According to the objectives, three work packages (WP) have been allocated. Proposed tasks, and expected outputs are explained in the 'case for support' document.
1)WP1 Conduct a scoping study to refine research questions and identify relevant resources: Step one will first explore the substantial inequalities dimensions from existing national datasets, and the psychological responses to the climate crisis in different groups. Step two will investigate available emerging digital data sources and their modelling and prediction techniques. Relevant research groups around the world will be recorded in a database in order to reach wider research communities and make contact with top academics in the future to simulate new research projects.

2)WP2 Develop an understanding of information requirements in climate decision-making: Climate information users' (e.g., water managers, housing allocation planners) interpretations of climate information may differ from what climate experts aim to communicate. A one-day hybrid workshop will be held with researchers, climate information providers and users to identify and understand the gap between the information communication of climate change and the requirements in decision making in climate practice and policy, the economic barriers and hidden costs caused by climate change's impact on wellbeing. The experience, best practice, and data interfaces in climate communication in the two countries will be compared and discussed, which will help to increase awareness and share effective interventions to respond to the psychosocial impacts of climate change.

3)WP3 Exchange researchers between the UK and South Korea and support ECRs: Staffs and ECRs from each side will conduct several short visits to each country, for research presentations, brainstorming and discussions. Especially, training programmes will be designed to support ECRs, which helps to build the long-term collaborations between the UK and South Korea groups. Remote discussions, meetings, and seminars will also be conducted via online platforms throughout the whole project period.

Publications

10 25 50
 
Description The objective of this project is to investigate the effect of climate change on psychosocial wellbeing inequalities with emerging sources of digital data, with the aim of simulating large research projects and fostering long-term collaborations. We have developed a research agenda that considers climate change and psychosocial wellbeing inequality within the same framework, identified priority research questions, and conducted interdisciplinary and collaborative research on larger projects globally. Through the scoping study and testing in specific research projects, we have:

1, investigated appropriate machine learning methods for effectively predicting multiple inequality outcomes from unstructured data, and evaluated the transferability of prediction models developed in data-rich settings to cases with limited data. For example, the LSTM is a suitable deep learning model that can accurately forecast economic indicators. Combining general and regional meteorological data can provide more accurate prediction outcomes. The accuracy of models like decision trees and random forests is largely improved when considering climate related variables.

2, evaluated the effectiveness and importance of climate related features in the prediction of economic indicators. For example, the weather impacts someone's mental state or disposition, preventing them from making rational or perfect decisions. People's decision-making is impacted by the weather, which may affect the movement of stock returns and volatility. For specific models such as Decision Tree and XGBoost models, CO2 emissions produced significant contributions to the prediction. The quantity of CO2 emissions can serve as an indicator of futures price movement. Sunshine duration also contributed to the Decision Tree and XGBoost models, even though its weight was far less than that of CO2 emissions. The humidity information is also a significant factor in futures price forecasting.
Exploitation Route For the policymakers and decision-makers, the research outcomes can help identify the factors that contribute to inequality indicators and make data-driven decisions to address the issue. Our findings on the effectiveness of climate-related features in predicting economic indicators could also inform policies related to climate change mitigation and adaptation.
Sectors Digital/Communication/Information Technologies (including Software)

URL https://www.linknetwork.online/
 
Description The project raised awareness of climate change's influence on psychosocial wellbeing, and highlighted issues where emerging digital data and machine learning may improve modelling and predicting such influence. With the support of this grant, we actively engaged and built collaborations and partnerships with policymakers, local authorities, and industrial and business partners such as the Leeds City Council, the Met Office, and regional companies. New and advanced data analytics techniques are developing during those collaboration projects. The research outputs can contribute to the development of new machine learning algorithms and data analysis techniques that can effectively predict multiple inequality outcomes from unstructured data. Such effective data reporting tools will help policymakers and the general public make informed decisions. The research outputs can also serve as a basis for further research in fields such as climate studies. Researchers in these fields can build on our findings and explore how climate-related features can affect inequality outcomes in different contexts.
Sector Digital/Communication/Information Technologies (including Software)
Impact Types Policy & public services

 
Title Tabular Image: a method to convert tabular data to images for convolutional neural networks 
Description We propose a novel data transformation method called Tabular Image that converts tabular data into images to take advantage of the powerful two-dimensional convolutional neural networks that perform extremely well on images while mitigating the challenges tabular data poses to deep networks (i.e. a mix of feature types and data sparsity). 
Type Of Material Data analysis technique 
Year Produced 2023 
Provided To Others? No  
Impact The Tabular Image provides a flexible framework that can be adjusted to suit various tabular datasets and requirements in other domains. Since tabular data is one of the most common data types in real-world applications and is widely used in climate science, medicine, finance, manufacturing, fraud detection, and many other applications that are based on relational databases, our proposed method may also help other domains take advantage of advanced 2D CNNs to improve the model performance in these domains further. 
 
Description Msc project with the Met Office 
Organisation Meteorological Office UK
Country United Kingdom 
Sector Academic/University 
PI Contribution Can we get an understanding of the public discourse around climate change, and how this is impacted by particular events? Particularly whether the sentiment changes, i.e. is it more positive or more negative? Based on such research questions, we have created a collaboration with the Met Office to create several Msc level student reserch projects.
Collaborator Contribution Co-supervise the students and provide relevant climate data
Impact Three Msc projects (London flooding July 2021, First exceedance of 40°C in UK, July 2022 and the COP26/27 November 2021/2022) on social media data research - sentiment in climate related tweets
Start Year 2023
 
Description Invited talk in International Symposium on Forecasting 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other audiences
Results and Impact Dr Jooyoung Jeon presented research "Monitoring & Probabilistic Forecasting of AsthmaMonitoring & Probabilistic Forecasting of Asthma" in the International Symposium of Forecasters, Oxford, UK
Year(s) Of Engagement Activity 2022
 
Description Invited talk in International Symposium on Forecasting 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other audiences
Results and Impact ECR Junhao Liang presented research 'a method to tabular data to images for convolutional neural networks' at the International Symposium on Forecasting, Oxford, UK, 2022
Year(s) Of Engagement Activity 2022
 
Description Mini workshop 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach Regional
Primary Audience Postgraduate students
Results and Impact A number of postgraduate students from the MSc Business Analytics & Decision Sciences programme presented their research on climate data analytics in a mini-workshop in collaboration with the Met Office and the Centre for Decision Research at Leeds University Business School. The best presentation award went to the research project on "perceptions and behavioral responses to the severe wind, rain, and snow weather warnings of UK residents based on longitudinal data'.
Year(s) Of Engagement Activity 2022