# Supply chain satellite monitoring of outdoor storage sites Satellite imagery shows what lies in the open at storage sites, ports and supplier yards: stockpiles, containers, tanks and vehicles. Repeated images turn this into time series of activity, and machine learning can detect and count objects automatically. Combined with shipping data and your own figures, this helps plan supply chains for sites you cannot visit. ## The task Supply-chain planning depends on what is happening at sites far away: how much material sits at a supplier's yard, how full a port's storage areas are, whether production at a plant is running. Reports from suppliers and partners can arrive late, and some sites are rarely visited. Much of this activity takes place in the open. Satellite imagery supports object counting for stockpiles at ports or silos, and 3D models support the quantification of inventories [1]. Commodity traders monitor goods production and shipping at key locations to predict supply and demand, and Earth observation contributes to this [1]. A company can apply the same approach to the sites its own supply chain depends on. ## Which data shows activity at storage sites | Data | What it shows | Example | | ----------------------------------- | ----------------------------------------------------- | ---------------------------------------------------- | | Very high resolution optical images | Single objects: stockpiles, containers, tanks, trucks | Pléiades Neo, 30 cm, revisit twice a day | | High resolution optical images | Larger changes in the area used for storage | Copernicus Sentinel-2, 10 m, revisit five days, free | | Radar images | Acquisitions through clouds and at night | Copernicus Sentinel-1 | | Shipping data | Vessel arrivals and departures at port sites | AIS position reports | | Your own data | Orders, deliveries and stock figures | ERP and supplier reports | Specifications from ESA for Pléiades Neo [6], Sentinel-2 [5] and Sentinel-1 [7]; shipping data as used in the PSVE project [3]. Copernicus data are free of charge [8]. **Figure:** Illustrative map, not real data. Outdoor storage sites of suppliers around a plant, as a storage-site analysis would mark them. ## How the analysis works 1. **Select the sites**: the supplier yards, ports and depots that matter for the planning decision. 2. **Collect images over time**: a series of acquisitions for each site, at the resolution the objects require. 3. **Detect and measure**: machine learning detects objects in the images and measures the area or volume in use. Research datasets for this task contain objects such as ships, storage tanks, harbours and large vehicles, which vary widely in scale, orientation and shape [4]. 4. **Combine**: the indicators are fused with other data. The PSVE project, for example, combined image classification with shipping data to estimate stockpile volumes at ports [3]. 5. **Feed the plan**: the indicators go into supply-chain planning as time series per site, next to the figures you already have. How AI turns images into information in general is the topic of our use case on [analysing satellite imagery with AI](https://leto.space/use-cases/ai-imagery-analysis/). ## Monitoring supplier sites The same images help watch the sites a supply chain depends on. Satellite Earth observation can monitor whether supplier operations comply with environmental standards [10], and ESA lists stockpile estimation among the applications for finance and insurance [9]. When a flood or a fire hits a region with suppliers, the Copernicus Emergency Management Service can map the affected area and the damage to infrastructure within hours or days, although only authorised users can request an activation [11]. The service is described in [rapid damage mapping after disasters](https://leto.space/use-cases/disaster-rapid-mapping/); how exposure to such hazards is assessed in advance, in [assessing climate risk for properties and assets](https://leto.space/use-cases/property-climate-risk/). ## What you get - a time series per site of the area or number of objects in storage - alerts when an indicator moves outside a range you define, for example a supplier's stock falling below the usual level - one view of all monitored sites, including sites no one from your company visits - a record of events such as floods or fires that affected supplier sites ## Limits **Limits** - **Only what lies in the open:** Satellites see stock that lies outside. Goods in warehouses, in closed containers or under roofs stay invisible, so the indicator describes outdoor storage only. - **Clouds hide optical images:** Optical images need a clear sky; radar satellites such as Sentinel-1 also acquire images through clouds and at night \[7]. A service that relies on optical images alone will have gaps. - **Estimates need calibration:** Counts and volumes from images are estimates. They become reliable for planning once they have been compared with known stock figures for the same sites. ## Our client project in this sector One of our four anonymised client projects covers this topic, in construction and real estate. It used Earth observation and AI. ### Client project in construction and real estate - **Topic:** Supply-chain planning through analysis of global outdoor storage sites. ## How we help Storage-site indicators are useful when they feed a planning decision. In [strategy and use cases](https://leto.space/consulting/strategy/) we check which sites and indicators answer your planning question and whether satellite data do this better than the reports you already receive. In [implementation](https://leto.space/consulting/implementation/) we run the pilot on a few sites and take the indicators into your planning routine. The marketplace on [Spacegraph](https://leto.space/spacegraph/), our discovery platform, has a use-case category for object detection and counting. ## Related use cases - [Monitoring global traffic hubs](https://leto.space/use-cases/traffic-hub-monitoring/): Ports and hubs that a supply chain depends on, from satellite and shipping data. - [Analysing satellite imagery with AI](https://leto.space/use-cases/ai-imagery-analysis/): How machine learning detects and counts objects in images. - [Construction and real estate](https://leto.space/industries/construction-and-real-estate/): All use cases for construction, infrastructure and real estate. - [Mobility and logistics](https://leto.space/industries/mobility-and-logistics/): Use cases for road, rail, aviation and maritime transport. ## Frequently asked questions ### Which image resolution do we need to analyse a storage site? It depends on what you need to tell apart. EUSPA notes that the resolution needed to distinguish assets on the ground differs widely between, for example, a large agricultural field and the stockpiles of an urban recycling site [2]. Free Sentinel-2 images have a resolution of 10 m [5]; commercial missions such as Pléiades Neo reach 30 cm [6]. The requirements should name the smallest object that matters for your decision. ### Can satellites measure the volume of a stockpile? Estimates are possible. The ESA-supported project PSVE set out to quantify the volume of dry bulk commodities such as iron ore and coal held at large international ports. It applies machine learning to Sentinel-2, PlanetScope and very high resolution imagery and fuses the results with shipping data; it passed its final review in 2023 [3]. Accuracy depends on the material, the site and the data, so a pilot should compare the estimates with known stock figures. ### How do we start? Choose a small set of sites that matter for one planning decision, for example the yards of two critical suppliers, and define the indicator, such as the area covered by stock. Then compare a few months of satellite indicators with the figures you already have. Our [consulting overview](https://leto.space/consulting/) shows the path from the problem to operation. ## Sources 1. [EU Space Market Report 2026, Issue 1, EU Agency for the Space Programme (EUSPA), 2026-05-26](https://www.euspa.europa.eu/sites/default/files/documents/EU_Space_Market_Report_2026.pdf) 2. [EUSPA EO and GNSS Market Report, Issue 2, 2024, EU Agency for the Space Programme (EUSPA), 2024](https://www.euspa.europa.eu/sites/default/files/external/publications/euspa_market_report_2024.pdf) 3. [PSVE: dry bulk commodity stockpile monitoring at ports (ESA BASS project), European Space Agency (ESA), ESA Space Solutions, 2023-06-22](https://business.esa.int/projects/psve) 4. [DOTA: A Large-scale Dataset for Object Detection in Aerial Images, Xia et al., IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2018, 2018](https://arxiv.org/abs/1711.10398) 5. [Sentinel-2 (mission page), European Space Agency (ESA), accessed 2026-10-06](https://www.esa.int/Applications/Observing_the_Earth/Copernicus/Sentinel-2) 6. [Pléiades Neo, European Space Agency (ESA), Earth Online, accessed 2026-10-06](https://earth.esa.int/eogateway/missions/pleiades-neo) 7. [Sentinel-1 mission… did you know?, European Space Agency (ESA), 2025-11-03](https://www.esa.int/Applications/Observing_the_Earth/Copernicus/Sentinel-1/Sentinel-1_mission_did_you_know) 8. [Copernicus, European Commission, DG Defence Industry and Space (EU Space portal), accessed 2026-10-07](https://eu-space.europa.eu/earth-observation/copernicus) 9. [ESA Space Solutions: Space for finance and insurance, European Space Agency (ESA), ESA Space Solutions, accessed 2026-10-06](https://business.esa.int/finance-and-insurance) 10. [ESA Space Solutions: Space for land mobility, European Space Agency (ESA), ESA Space Solutions, accessed 2026-10-06](https://business.esa.int/mobility-land) 11. [About the Copernicus EMS On-Demand Mapping service, Copernicus Emergency Management Service, accessed 2026-10-07](https://mapping.emergency.copernicus.eu/about/) Page: