Earth-orbiting satellites have been helping humanity to make sense of our evolving Planet for over fifty years. This high ground advantage has already helped to deliver insights to diverse audiences, from fisheries to geologists and governments, influencing the flow of trillions of euro in annual economic value and trade.
Demographics and migration are a key driver of socio-economic trends and security. Migration studies is the term given to this interdisciplinary study of the causes, processes and outcomes of human migration, and peer reviewed papers are published in leading journals including Migration Studies (Oxford University Press). This supporting research forms a basis for applying systems thinking to assist in the development of frameworks and tools to understand and forecast migration trends.
Stories of human migration have been recorded since the earliest civilisations. By studying historical records, researchers can build a picture of the causes, processes and outcomes. Migration scholars are able to reconstruct some of the macro and micro details of mass migration from data sources including religious texts, ships logs, birth certificates, and census documents.
Modern technology, and satellite sensors and machine learning specifically, offer a means of monitoring migration, and its causes, in near real-time. Especially in the case of forced migration. This insight is of value to governments, humanitarian agencies, NGO's and other stakeholders. UNHCR, the refugee agency, estimates that there are 68.5 million forcibly displaced people worldwide today. The following are four examples of how we can harness satellite data to build this picture today.
Refugee camps: identification and growth
Using imagery data from the optical sensors aboard satellites we can train machine learning models to identify temporary settlements at scale and track their growth and evolution over time.
Refugee camp near Shamarin, Syria (Credit: Digital Globe/Google)
Refugee camps: planning and maintenance
A recent collaboration between UNHCR (USA) and Stanford is seeking to build a model to tag and categorise the structure of aid camps themselves in an effort to streamline the planning and maintenance workload on teams.
Drought and famine are a primary driver of migration as those with the means to do so will move to safety and security in the event of food scarcity. The downstream impact of crop failure can be widespread social unrest, conflict, and political instability. Satellite imagery, multispectral and SAR (radar) datasets can be combined with non-Space data to build a realtime picture of agricultural conditions and output. Regional and local models of crop health can be trained with historical data to give an indication of future trends and risks.
Sentinel 1 / Copernicus (Credit: European Space Agency)
Similar to the previous example, we can train models to detect and categorise objects in the maritime environment that have similar characteristics ie. boat length. Depending on the location (latitude/longitude) of the area of interest the refresh rate may be longer than desired as we wait for the satellite to pass overhead again. Layering other datasets (ie. weather, tide, wake length, wave height, heading, agency) can help to build a predict of where an object may be on the next pass.
Limitations and next steps
While cloud coverage, resolution, satellite orbits and data processing resources all present challenges for accessing insights from Space, applied in the right context it can prove a powerful tool and a complement for other methods and approaches for making data-driven decisions. If you would like to discuss a problem or project in more detail, simply click on the chat icon in the bottom-right corner of this page to speak directly with the product team at thirdrock.io.