A Data-Driven Approach to Identifying Optimal Direct Air Capture Deployment Sites in Germany

Direct Air Capture (DAC) is a carbon dioxide removal (CDR) technology that extracts atmospheric CO₂ for storage or utilization. This study presents a data-driven approach to identify optimal DAC deployment sites in Germany, focusing on the North German Basin (NGB), a region with promising geological CO₂ storage capacity. Building on 92 identified storage traps, we integrate K-means clustering machine learning algorithm to evaluate surface conditions using both environmental and infrastructural datasets.

Criteria were selected from the literature and divided into onshore and offshore constraints, including factors such as population density, protected areas, pipeline access, seismic activity, and shipping routes. Spatial data processing involved aggregating raster inputs into 100 m² grid cells and calculating distance maps from key infrastructure and risk zones. All the features were homogenized in the same format to process it through K-means algorithm.

K-means clustering, supported by the Elbow method, grouped potential DAC sites based on geospatial similarities. Cluster results were compared using a 5-level suitability ranking system (J-scores), which assigned scores based on site characteristics and normalized them for comparability. The resulting surface suitability map highlights high-priority regions. The developed methodology can then be used for larger geological regions for which 5-level ranking suitability ranking system is not feasible. This methodology offers a scalable, machine learning-supported framework for DAC site selection that integrates geospatial, environmental, and infrastructure criteria. It serves as a valuable tool for planners, industrial operators and stakeholders aiming to strategically implement DAC. The method is transferable to other regions with appropriate data availability.