Using Satellite Data and Machine Learning to Study Conflict-Induced Environmental and Socioeconomic Destruction in Data-Poor Conflict Areas: The Case of the Rakhine Conflict

Dr

Roman Vakulchuk

(He/Him)

Senior Research Fellow

Faculty of Economics, Business & Management

Nord University

Roman Vakulchuk is a Senior Researcher at Norwegian Institute of International Affairs (NUPI) in Oslo, Norway. He is also Adjunct Professor and Researcher at Business School of Nord University in Norway, with a particular focus on scenario methodology. He is co-founder of Central Asian Development Institute.

Overview

The idea behind this article was to apply new, digital technologies in order to study some of the least accessible places in the world, namely neighboring regions of Bangladesh and Myanmar. We studied changes over time in environmental, social and economic domains, and explored the causes of these changes (2012-2019).

The purpose of this research was to find new ways to do research in areas that are difficult to reach, and therefore research with traditional methods such as interviews, survey, field work.

In our case, we looked at regions where the Rakhine people are located. The Rakhine people are an ethnic group with whom the central government in Myanmar has had a series of conflicts and tensions, and what little we knew based on some individual accounts of people who had been there and some limited journalistic accounts. This was not enough to make reliable conclusions about what has been happening in the area, and we wanted to explore the change over time in environmental, social and economic domains.

Key Findings

While there are many indicators used to estimate the social, environmental and economic situations of areas, we need new indicators to analyse satellite imagery.
The 2017 conflict has led to impoverishment, deforestation and an overall worsening of the situation in those regions. Our indicators showed that the range of household income is lower, and deforestation has strongly increased.
From 2012 to 2019, the size of forests in the bordering regions on the side of Myanmar and Bangladesh decreased by 20% and 13% respectively.

    How to apply research

    The international aid community that works on Myanmar can use the evidence we've found in negotiations, in designing development aid projects, in gaining a better understanding of the situation on the ground.
    Our study encourages researchers to expand their methodological tools and to try to do research with a larger variety of methods, with more innovative tools, in order to gain a more nuanced understanding of complex situations. Researchers can sometimes tend to do research the way they've been taught, but academic research would benefit from more risks being taken and more innovative techniques being used.

      Let your research make a social impact

      About this research

      This journal article was part of a collaborative effort

      Thiri Shwesin Aung

      Indra Overland

      Yanhua Xie

      This research was independently conducted and did not receive funding from outside of the university.

      Recommended for

      UN Sustainable Development Goals

      This research contributes to the following SDGs

      About this research

      This research was independently conducted and did not receive funding from outside of the university.

      This paper was co-authored

      Thiri Shwesin Aung

      Indra Overland

      Yanhua Xie

      Recommended for

      What findings means

      This research has yielded a couple of findings:

      First, we found that the methodology we used still needed to be developed, especially in terms of the kind of indicators that can be used to draw conclusions. Many indicators can be used to estimate the social, environmental and economic situations of the areas, and it is important to know which to use where (for instance, one can estimate the economic state of an area by looking at the state of roofs on houses). It is important that future research explores in more detail these indicators and finds new ones.

      Second, we found that the economic, social and environmental situation has largely worsened. Our indicators showed that the range of household income is lower, and deforestation has strongly increased. What we concluded was that the main driver of these tendencies was the 2017 conflict, which has led to impoverishment, deforestation and an overall worsening of the situation in those regions.

      Methodology

      We analysed Google Data Storage imagery (high resolution satellite imagery of bordering areas between Myanmar and Bangladesh where Rakhine people live) with machine learning algorithms in order to assess the change over time (from 2012 to 2019) in terms of the environmental, economic and social state of the areas. We used a variety of indicators, from the quantity of trees to the state of roofs on houses.

      However, it should be remembered that our research denotes very general dynamics, and we cannot explain specific developments like, for instance, why 5% of hospitals in the areas have different roofs. We can make general conclusions about large-scale developments, but our analysis cannot give us answers for more specific findings.

      Another limitations is the fact that our research could be difficult to replicate or do on other areas, because obtaining the satellite imagery we used was expensive. Smaller-scale researchers will not have the access to this equipment.

      Glossary

      Fragility, conflict and violence
      This concept was our theoretical framework, our frame of reference for the theoretical approach we took to the research. It is a concept used to describe settings which are extremely fragile and are very prone to conflict and violence. This concept also refers to the fact that these areas need to be treated very carefully and with nuance, be it in research, national assistance, aid.

      Want to read the full paper? It is available open access

      Aung, S. T., Overland, I., Vakulchuk, R. (2021). Using Satellite Data and Machine Learning to Study Conflict-Induced Environmental and Socioeconomic Destruction in Data-Poor Conflict Areas: The Case of the Rakhine Conflict. Environmental Research Communications, 3, 1-18.

       

      Thank you to

      for helping to prepare this research