The alarming loss of forests worldwide has devastating consequences for biodiversity1, ecosystems, and human livelihoods, intensifying climate change and creating a critical global challenge that demands urgent solutions. The ambitious goal to end deforestation by 2030 was specifically put forward at the United Nations COP26 climate summit in Glasgow in 2021. While global deforestation continues, its rate has slowed over the past decade, particularly in South America, where losses halve compared to 2000–2010. Yet this progress remains insufficient to achieve net-zero deforestation, as Africa’s accelerating losses and persistent global forest decline demonstrate2. Critically, these trends are driven by unsustainable global consumption: agricultural expansion alone accounts for 90% of forest loss, demanding urgent, supply-chain-focused interventions3,4.

Since the European Union (EU) is a major consumer of commodities associated with deforestation, the EU announced Regulation (EU) No. 2023/1115 for Deforestation-Free Products5 (hereafter EUDR), which entered into force on 29 June 2023. As regulation with global scope, it aims to ensure that commodities, namely cattle, cocoa, coffee, oil palm, rubber, soya, and wood (Fig. 1) and derived products, such as leather, chocolate, or furniture, placed on the EU market, regardless of their origin, have not contributed to deforestation or forest degradation after December 31, 2020 (cut-off date). The EUDR mandates that non-SME (small and medium-sized enterprises) operators and traders conduct due diligence in their supply chains to prove that commodities are not linked to deforestation, forest degradation, or illegal practices. This involves collecting information, assessing risks, and submitting due diligence statements. Nonetheless, the regulation remains contested and is subject to ongoing revision by the European Commission, with implementation delayed until 30 December 2025 for large companies and June 2026 for SMEs6, which have fewer obligations than non-SMEs. Recent simplifications (April 2025) aim to ease compliance, though further adjustments are likely given evolving global dynamics.

Fig. 1: Mapping land use following deforestation.
figure 1

Global observing systems for EUDR-compliant monitoring of soya, cattle (indicated as pasture), palm oil, wood (indicated as plantation forest), rubber, coffee, and cocoa plantings at selected hot spot sites, shown in RGB (Google Earth Engine base maps) and classified formats. Note that the images are conceptual representations and not exact depictions of the actual locations. EO data source for classified images was high-resolution Planet-NICFI (4.77 m), and the methodology for processing the maps was based on deep learning models, see details in Masolole et al.14 (Credits: Markus Immitzer and Talita Nogueira Terra Parizzi for photographs and Robert Masolele for https://robertnag82.users.earthengine.app/view/africalu).

Full size image

Here, we argue that mandating EO-based monitoring under the EUDR is critical to verify supply chains, ensure deforestation-free compliance, and assess environmental impacts. A mandatory approach mitigates key risks, such as inconsistent enforcement (e.g., due to uneven national capacities), fragmented data sources, and the lack of scalable in situ proxies.

EO’s role in addressing EUDR implementation

EO serves as the foundational infrastructure for assessing EUDR compliance by enabling transparent, near-real-time monitoring of deforestation risks and commodity expansion across global supply chains7.

The Sentinel-1 and Sentinel-2 satellites—jointly developed and operated by the European Space Agency (ESA) and the European Commission—alongside NASA’s Landsat series, form the backbone of operational forest change monitoring due to their global coverage, open data policies, and suitability for long-term trend analysis8. Landsat enabled the creation of thematic products, such as the 30 m-resolution global forest change (GFC)9,10, dataset, which tracks forest loss and gain since 2000 and serves as a baseline for many systems, e.g., the global land analysis and discovery (GLAD-L)10. However, Landsat’s 30 m spatial resolution can limit the detection of small-scale deforestation or subtle forest degradation. GLAD-S2 extends the system to Sentinel-2, providing near-real-time detection of primary forest loss at 10 m resolution. The radar for detecting deforestation (RADD) is a near-real-time radar-based (Sentinel-1) alert system for the tropics provided by Global Forest Watch (GFW)11. GFW recently introduced DIST-ALERT, which significantly expands monitoring capabilities, but with its 30 m resolution, the system is limited in reliably detecting small-scale disturbances (e.g., selective logging or narrow clearings) that are visible only at higher resolutions. Additionally, the ≥30% vegetation loss threshold means that subtle degradation events or disturbances in sparse-canopy forests (10–30% cover) may be missed.

Regarding EO for agricultural land use monitoring of EUDR relevant commodities, Sentinel-2 has proven its abilities in monitoring large-scale monoculture crops like oil palm12 and rubber13, but for monitoring small-scale or agroforestry systems like coffee and cocoa, optical very high-resolution (VHR) imagery with a pixel size <5 m offers better monitoring performance14.

The integration of multiple datasets15,16 empowered initiatives like the Food and Agriculture Organization of the United Nations (FAO) “What is in that plot?” (WHISP)17, an open-source geospatial tool that applies a convergence-of-evidence approach for policy-ready analysis. WHISP is among the most advanced free and open-source tools for preliminary EUDR due diligence, offering cross-validation of multiple datasets (e.g., GFC, RADD, Global Forest Cover map 2020 - GFC2020 v2, Tropical Moist Forest -TMF, crop maps) to address compliance needs. However, its current scope is limited to humid tropical regions, lacks integration with VHR imagery for small-scale deforestation or farming detection, and is less flexible and customizable than commercial systems.

Multi-sensor European systems for multiple purposes

The situation of EO capability will improve in the early 2030s with Sentinel-1 NextGeneration (NG) and Sentinel-2NG, providing higher temporal, enhanced spectral and better spatial resolutions compared to the current systems. This will specifically enhance the ability to monitor dynamic land use practices and smaller-scale changes. In addition, ESA plans the launch of several expansion missions within Copernicus, a long-term, fully operational EO program, from 2028 onwards. These include the Copernicus Hyperspectral Imaging Mission for the Environment (CHIME), the Land Surface Temperature Monitoring (LSTM) and the Radar Observing System for Europe in L-band (ROSE-L).

Hyperspectral data complements commodity and forest monitoring (see Fig. 2), allowing for instance improved assessment of agricultural crop characteristics and residues18, as well as tree species distribution19, supporting the distinction of primary, naturally regenerating and planted forests. High spatial-temporal resolution thermal sensors will allow fine-tuning of land cover and vegetation mapping, useful for improved differentiation of commodities20. Longwave radar all-weather imagers can sense heavily vegetated areas such as the tropical rainforests, thus strongly supporting due diligence with forest mapping and biomass monitoring in one of the most threatened biomes21. All this new global data will come with a free and open data policy.

Fig. 2: Use of EO data/products for different EUDR stakeholders.
figure 2

Multiple stakeholders are affected by the EUDR and can make use of EO data and products for different purposes. The stakeholders and associated purposes with high, medium and little / no relevance were derived from the authors’ common understanding and supported by references cited in the commentary. Certifiers: note that external certification currently does not substitute the operator’s responsibility regarding due diligence. Exemplary future key missions (except for Sentinels, only spectral domains are mentioned to keep it general) for specific purposes are given.

Full size image

EO data have been primarily used to map land cover and use, including deforestation, cropland extent and type, a non-trivial task fundamental to agricultural and forest monitoring. Beyond this baseline, the new generation EO data can also be explored to estimate advanced crop variables like yield22, biomass23, and productivity24, which may enable a more comprehensive environmental and economic assessment in the context of the EUDR12,25.

Thus, for future efforts, we propose a multi-sensor monitoring system for multiple purposes, ranging from understanding commodity expansion, deforestation trends and environmental impacts, to risk assessment, general transparency and traceability of supply chains (see Fig. 2).

Loopholes of the EUDR

There are ongoing criticisms, debates, and identified loopholes regarding the EUDR. One key discussion is that producers may redirect compliant goods to the EU while continuing deforestation for other markets26. Another possible issue is misreporting, where commodities are claimed to originate from compliant regions but are sourced from deforested areas.

A growing concern is ‘cattle laundering,’’ a deceptive practice where illegally raised cattle from deforested land are mixed into legitimate supply chains. While EO data (e.g., Sentinel-2) can detect deforestation, e.g., at rearing locations, it cannot track individual animal movements. This gap creates opportunities for fraud, but blockchain technology could address it by documenting each transfer between farms, pastures, and slaughterhouses27. Hence, combining EO alerts to identify deforestation hotspots, and blockchain reveals hidden transactions, even when cattle are moved through ‘clean’’ intermediary farms.

Supporting this argument, a study28 found 69% of post-2011 deforestation shifted to smallholders, groups often being excluded from traditional traceability systems. Here, EO plays a critical role in identifying high-risk zones, while blockchain tracks commodities through complex supply chains to meet EUDR due diligence requirements.

Effective data governance requires interoperable systems between blockchain and EO platforms. The EU should mandate standardized open APIs to enable automated cross-verification of satellite-derived deforestation alerts (e.g., from Sentinel-2) with on-chain supply chain records. This integrated approach would address critical cattle-tracing gaps under the EUDR while maintaining data integrity and privacy.

By adopting such technology, combined with Artificial Intelligence (AI) and EO, the compliance costs associated with the EUDR may even fall well below the recently estimated average of only 0.1% of annual revenues29 for large companies and SMEs. European companies are likely to pass compliance costs on to primary producers. These requirements will apply universally, even to those already operating deforestation-free. While large producers or firms in advanced developing countries can absorb these costs, small producers in poorer regions face significant market barriers, even with EO tools. Moreover, the recent adjustments to environmental and land tenure laws in countries, made to comply with the EUDR, are criticized by environmentalists as they may lead to a weakening of domestic forest protection efforts30.

Scientific studies31 have highlighted critical lessons learned from the Reducing Emissions from Deforestation and forest Degradation (REDD)+ program32, which has evolved to recognize the complex realities of forests, including social, environmental, and economic factors. The specific scope of the EUDR can lead to oversimplifying the issue, focusing mainly on the conversion of forests to agriculture and conversion to plantations, while ignoring other important changes happening within the forest-agriculture landscape. For instance, the risk of misclassifying agroforestry systems as forests will potentially harm sustainable agro-practices, as these are considered ‘deforestation-free’ if in place before the cut-off date31,33. The EUDR adopts the FAO forest definition, which uses fixed thresholds (0.5 ha minimum area, 10% tree cover, and >5 m height)2. These parameters create monitoring challenges, as the 0.5 ha minimum area threshold falls below the reliable detection limits of Landsat (30 m) and Sentinel-2 (10 m) data8. Furthermore, current EO systems can monitor trees outside forests—a capability that could help avoid underestimating tree cover loss in fragmented landscapes and small-scale clearings8. These discrepancies highlight the need for regulatory frameworks to evolve in tandem with advancing EO monitoring capabilities.

A study34 demonstrated that voluntary zero-deforestation commitments often fail due to inconsistent monitoring, particularly where certification programs lack near-real-time EO verification (e.g., outside the Brazilian Amazon). Their finding that 50% of commitments allow compensatory reforestation or delayed implementation underscores why the EUDR must mandate EO data: only systematic satellite monitoring can provide the timely, standardized evidence needed to close these loopholes (Box 1).

Bridging EO opportunities and stakeholder needs

Given the EU’s leadership in space science35 and its growing Copernicus capabilities, users worldwide benefit from free, open, operational, and long-term satellite data, offering critical stability and reliability for EO-based regulatory systems. In the context of the EUDR, multiple stakeholders (Fig. 2) face distinct challenges in utilizing EO. Regulators require standardized, near-real-time deforestation alerts (e.g., RADD, GLAD-S2) to verify compliance (i.e., competent national authorities) and enforce risk-based benchmarking6 (i.e., European Commission), yet gaps persist in detecting small-scale clearing. Commodity suppliers (traders and operators) need cost-effective, scalable tools (e.g., WHISP) to demonstrate due diligence, but may not be familiar or aware of the tools. Certifiers, although currently not substituting for the diligence of operators, may play a crucial role in independently verifying EUDR compliance, using freely available EO data and products (e.g., Sentinel-2, GFC) to screen for deforestation risks in supply chains. However, they may lack access to VHR imagery (e.g., Planet) needed, for instance, to resolve small-scale degradation or to differentiate agroforestry from forests7, which can lead to uncertainties in contested cases. Exporting countries may face challenges in integrating EO data (e.g., Sentinel-2 alerts) with national systems, such as forest inventories, due to technical gaps or outdated infrastructure. Indigenous communities may provide critical ground-truth data to train and validate EO-based deforestation alerts (e.g., confirming illegal encroachments flagged by Sentinel-2), enhancing detection accuracy for small-scale clearing that satellites alone may miss. Consumers and researchers increasingly demand transparency in product origins to assess climate impacts, driving adoption of public EO platforms (e.g., from GFW) that map deforestation risks to supply chains in near-real-time.

The untapped potential of EO

EO can address critical EUDR implementation gaps from detecting deforestation to monitoring crop-driven leakage in a rapid and transparent manner. Yet, the success hinges on overcoming some remaining technical and governance challenges. The EU should fund training programs for non-EU producers to ensure equal access to EO tools and reduce compliance asymmetries. Further, to ensure fair and effective monitoring, the EU should create joint teams with producer countries and local communities. These could work together to design deforestation tracking systems that respect local needs and protect small farmers from being unfairly excluded. Hence, with the upcoming EUDR implementation phase approaching in 2026, scaling EO’s role requires implementing AI-driven near-real-time monitoring.

To fully adopt EO for EUDR enforcement, several developments are critical: (i) validation protocols that ensure consistent verification of deforestation; (ii) open training datasets to enhance methods for detecting land use change; and (iii) independent monitoring systems that link EO alerts with customs and land registry data. The Common Agriculture Policy monitoring system, combining satellite checks with sampled field inspections, provides a proven model for cost-effectively scaling EO verification.

We acknowledge that the regulation’s evolving timeline6 may require adaptive monitoring approaches. Therefore, future research should assess how (future) EO systems can remain effective given ongoing policy revisions and phased rollouts.

Institutionalizing EO in a multi-sensor system would thus enhance long-term regulatory effectiveness by ensuring standardization, transparency, and equitable enforcement across the EU market. By integrating satellite technology innovations with stakeholder engagement, EO can become a cornerstone of the EU’s deforestation strategy, strengthening regulatory accountability while supporting global climate change mitigation.