Predicting forced migration from Copernicus environmental data, thus softening the negative impact of population movement.

  • 5,558 Raised
  • 1 Judges


  • Estonia


  • 3. Understanding and forecasting forced migration


Our team


Every year millions of people are displaced either by force or circumstance; either way involuntarily. This does not only affect the people displaced, but also everyone in the world. Predicting future displacements would help avert migration disasters by letting destination areas prepare in a timely manner.


While migration prediction models that aim to predict how human mobility is driven by economy, geopolitics  and environment metrics have come a long way they still are not precise in understanding human movement.  As climate change will force more and more people to migrate to more sustainable countries it is imperative that we understand  even the minute reasons behind emigration.


To leverage the Copernicus Data Space Ecosystem as well as data from EUROSTAT, World Bank, The Humanitarian Data Exchange, and other climate resources to develop a working model that can predict human movement even in peace times.


The goal is to leverage climate and environmental data from Copernicus Data Space Ecosystem to understand what are the potential driving factors for climate migration. Knowledge extracted from these data will help preemptive preparations for humanitarian aid and in any other possible way. We can enrich the data by using other sources from already existing organizations.


Being able to predict years ahead of time how many people are immigrating from a country will help to understand the movement of people and also give a chance for humanitarian aid to be planned ahead of time. 

Working together with governments we could enact policies that help people around the world. While similar products already exist in the market they tend to be tailored towards specific countries. 

Our advantage is that we will use the cutting edge Copernicus and sentinel data, and the goal would be to apply it in any part of the world. The side benefit would mean market prediction of export, food production, fresh water supply and housing.

Overlapping projects and organizations in the world


The challenge, among others, is to decide which metrics to choose from the massive store of Copernicus data. We can join together the knowledge and experience of people from several different domains. The end goal would be to create a platform that would connect data to institutions that it would provide service for and be used as an argumentative usable tool for future policy.

We need to comprehend what kind of data can predict human migration. Useful starting point are vegetation index, are of water bodies, general temperature, wind, sea levels, algae growth. And make them into a workable dataset to pinpoint ahead of time where and how many people will move from. Perhaps even to estimate how many people will not be able to live in certain places due to these factors. And then seeks ways to directly help those regions and people.

Initial plan for the Hackathon

  • Gaining the understanding what indicators and datasets to use
  • Streamlining the data capture from sources
  • Building the necessary virtual infrastructure
  • Leveraging machine learning for our purposes
  • Knowing where data enrichment with outside sources is necessary
  • Mapping out the field of competitors and collaborators
  • Developing a business Model
  • Developing a prototype to show

Sources to our Hackathon work

The working example of a website that users would leverage to gain knowledge about the possible upcoming crises shows us the countries of the world with necessary data points that might be used for our predictions.

The code repository can be found in github. We found value in a The Humanitarian Data Exchange website as a real life example that visualizes drought in the Horn of Africa region as an example.

Data comes mainly from the Copernicus satellite (using OpenEO). We use mainly NDVI (Normalized Difference Vegetation Index) ja NDWI (Normalized Difference Water Index). These can be used to map the local vegetation and water bodies from the back reflective light spectra. 

By analyzing migration data by gender, year, destination, and source country, we can leverage vegetation and water body data to predict migration and thus help with resource management and anticipate disastrous migration events. 

Some of the problems that arose during the Hackathon were that the satellite datasets are large and it’s slow to get aggregate measures. The queries run for an unknown amount of time, with large queue waiting times and sometimes fail with error messages that are difficult for the non-specialist to understand. Migration data is not on the same temporal and spatial scale as the satellite data– i.e. it’s very coarse-grained (yearly data by country). Finer grained data that is not currently available would enable more insights and better and more useful predictions. One of the functionalities that would add value to the Copernicus experience, would be to have an estimate of data set size/run time.


Our statistical analysis revealed that regions most impacted by drought did not always overlap with the greatest human burden. Using satellite data we found evidence that the drought didn’t start in 2019 as previously thought, but probably earlier. Such insight is valuable to decision makers as it is estimated that 7 more billion dollars will need to be invested to address the crisis, in addition to 2.6 billion already spent, not even addressing the cost of future crises that may have a similar environmental origin. This use of scientific earth observation to reduce sources of uncertainty in humanitarian crises is our signature and also what differentiates us from similar service providers, for example Danish Refugees Council, Save the Children ja UNHCR (United Nations High Commissioner for Refugees).

The estimated current market size for migration prediction stood at $4 billion in 2022, and is expected to reach 8.2 billion by year 2028. Our main source of funding will be from government grants. (Last year, 600 millions USA and $1,2 billion world-wide).

Our project document