1. Introduction: AI for Precision in Humanitarian Response

The International Committee of the Red Cross (ICRC) developed an Artificial Intelligence model and a data platform in partnership with leading engineering universities in Switzerland to access precise and up-to-date population data for effective humanitarian planning and response. The AI model, called POMELO (Population Mapping by Estimation of Local Occupancy Rates), provides fine-grained population maps to establish the size and density of populations, as well as associated information such as settlement type and population trends at a resolution of 100 x 100 meters. While facing technical challenges to operationalize the model, the ICRC developed a platform to guide users in the selection of adequate population datasets. The platform boosts ICRC’s capacity to plan and execute well-targeted humanitarian aid, allocate resources efficiently, and improve crisis response, ultimately ensuring more precise and impactful humanitarian interventions.

2. Background: The Critical Challenge of Missing Population Data

The lack of accurate and up-to-date population data (census) is a critical challenge in planning and responding effectively to crises. Often data is outdated, not granular enough, or just not available at all due to conflict, instability, or other reasons. Humanitarians traditionally use estimation methods such as counting roofs in satellite images and multiplying them with the average size of a household in the country. However, as Thao Ton-That Whelan, a GIS and remote sensing analyst at the ICRC, states: “The roof counting is not enough to have a good estimation of the population because it’s only a roof. You must know what type of roof it is. If it’s an administrative building, school, industrial building, or if it’s a residential building, and if it’s a residential building, is it a one-story or two-story building.” To achieve much higher precision in population estimates, ICRC looked at how additional and more detailed data points might be used, and thus POMELO was born.

3. AI Development: POMELO – A Machine Learning Approach to Population Mapping

Mapping and remote sensing through satellite imagery are methods widely used by non-profits to identify areas of intervention and plan operations as accurately as possible. There are multiple data sources that provide large amounts of data, yet resources in data analytics teams are scarce. Thao Ton-That Whelan has therefore been looking at ways to employ AI to automate or semi-automate processing, combining and interpreting data, so that her team can focus on in-depth analysis. Additionally, she explored how machine learning might be used for the extrapolation of information not directly available.

POMELO is one example of this work. To build it, ICRC collaborated with the Swiss Federal Institute of Technology Lausanne (EPFL) and the Swiss Federal Institute of Technology Zurich (ETHZ), who brought their expertise in Machine Learning and AI development to the project.

To ensure the model’s accuracy, extensive testing was conducted across several countries in sub-Saharan Africa. By comparing POMELO’s population maps to the most detailed existing population counts, the team found that POMELO’s predictions were more precise than traditional estimation methods.

4. Operationalizing AI: The Population Grid Hub

The ICRC faced technical hurdles to implement the AI model past its development phase by academic collaborators. Using POMELO requires a level of technical expertise challenging to develop and maintain internally. However, the development of the AI model motivated the ICRC to create an internal Population Grid Hub portal. Here, field staff can easily access and analyze population settlements and density in various regions. The portal homepage provides information about the tool, its purpose, and how it can be used. The hub allows staff members to zoom into specific areas and draw the outline of a geographic area for which they need population data.

The Hub also provides access to freely available population data so that users on the ground may validate the output. Ultimately, it is the user — with their knowledge of the local context — who decides which source gives the most reliable data for their objectives. User feedback has been positive, particularly noting the time saved in establishing population data; for selected locations, the accuracy of the algorithm’s output was measured and found to be good.

To further promote internal use, ICRC has prepared comprehensive documentation and organized webinars for staff members. While the Hub is managed and used internally by ICRC, the POMELO model code is publicly available on GitHub and can be used by other organizations grappling with similar issues in estimating population numbers.

5. Impact on International Development, Humanitarian Action, and Monitoring & Evaluation

6. Key Learnings: AI Sustainability and Organizational Capacity

While not all project objectives were reached regarding the full implementation of POMELO, the main outcome was successful: providing the tools for data-informed decision making through the data platform. The Population Grid Hub now serves as a reference point for field staff to validate and access population estimates, even if the underlying AI model requires specialized maintenance.

7. Future Directions: Detecting Displacement with Nightlights AI

In a related project, Thao Ton-That Whelan is looking to find ways of establishing population movements — something POMELO is not able to do with current static data inputs. By analyzing daily images of night lights provided by NASA, she hopes to find indications of both population movements and infrastructure damage. Nightlights can be taken as a proxy for human presence; when they are no longer visible, it may indicate damage (e.g., to a power plant), which would have knock-on effects on people, healthcare, and the need for intervention on the ground. This next-generation AI application could transform how humanitarian actors detect displacement in near real-time, enabling earlier response and more effective M&E of displacement trends.

8. Frequently Asked Questions (FAQ)