Source: French to English Tester Published on: 2026-05-21
Source: The Conversation – France (in French)– By Carl Bethuel, Doctor of Geography, University Rennes 2 It is one of the paradoxes of our time. Satellites, planes, drones, and artificial intelligences allow us to map the world with unprecedented precision.
Yet, numerous maps of the same place show different realities. How can this be explained, and how can one avoid falling into a “map-washing”where each person would take the card that suits them?” Never before have we mapped the planet so extensively.
Driven by the rise of sensors, computing platforms, and artificial intelligence, we now have an abundance of data thatprofoundly changes our way of observing the Earth. This revolution relies on a central tool: theremote sensing, which consists of observing the Earth’s surface remotely thanks to images acquired by satellites, airplanes, or drones.
Where cartography was once slow and localized, it is now possible to monitor territories on a regional or even global scale at an unprecedented pace.
A revolution in the service of major environmental challenges This ability to observe the Earth is notably due tomajor environmental issues, largely registered in theSustainable Development Goals (SDGs)of the United Nations, such as fighting climate change, preserving biodiversity, or the sustainable management of land.
Having precise mapsis essential to monitor changes, guide public policies andstrengthen the transparency of agricultural supply chains. This dynamic is currently driven by a multiplication of actors producing and mobilizing maps: public institutions, international organizations, researchers, non-governmental organizations, but also private companies developing their own satellite monitoring services.
This diversity is reflected in regulatory measures such as the European regulation against deforestation and forest degradation, which requires companies to demonstrate thatcertain raw materials do not contribute to deforestation. Its implementation relies heavily on the use of cards mobilized at different stages of the supply chains.
The case of oil palms: too many maps, not enough clarity In this context, the case of oil palm plantations in Indonesia is at the heart of concerns related toconversion of tropical forests into agricultural monocultures.
The world’s leading producer since 2007, the country faces challenges such ashabitat fragmentation, thedecrease in biodiversityand thedegradation of forest carbon stocks. The expansion of plantations is also accompanied by socio-economic tensions, particularly aroundland rights of indigenous peoples.
Many cartographic products have thus been developed to locate plantations and monitor their evolution, using various data and methods. This diversity offers a plurality of perspectives on the same object of study, constituting a scientific wealth.
However, it is accompanied by results that are sometimes difficult to compare.
Thus, four maps of oil palm plantations on the island of Sumatra yielded four different measurements of the extent of these plantations, with a difference of sometimes almost 2 million hectares, which is more than the area of Brittany.
Different mapping methods lead to different surface area calculations For example, the resolution of the images used or the observation period can vary, influencing the estimated areas. Differences in the definition of what is being mapped can also generate different results.
In the case of oil palm plantations, an apparently simple question actually proves complex: what exactly is included? Depending on the approaches, it may involve only industrial plantations or also include small peasant farms. Added to this is the question of temporality: a mature plantation, an immature one, or a plot in transition will not always be represented in the same way.
For end users (public decision-makers, NGOs, or private actors), this diversity of methods and definitions can become difficult to interpret. Faced with this profusion, one may wonder: which information to use? And how to compare results based on different hypotheses?
By multiplying the representations of the same phenomenon, the information becomes more complex, especially when uncertainties are not very visible. Too many cards: how to see more clearly? Faced with this multiplication of maps, several avenues are emerging to improve their use and comparability.
The challenge is no longer just to produce more information, but to strengthen its robustness and readability. Information fusion appears as a key approach.
Rather than considering each map as an independent truth, it is possible to combine them, like multiple viewpoints on the same phenomenon, in order to obtain a more stable vision.This is the approach we proposeas part of the ANR PALMEXPAND project (ANR-20-CE03-0004), conducted by an interdisciplinary team (CNRS, Cirad, Inrae).
This approach relies on Dempster-Shafer theory, which allows for combining multiple sources while taking into account their agreements, disagreements, and uncertainties. DEMPSTER-SHAFER THEORY Dempster-Shafer theory is a probabilistic mathematical approach developed in information sciences to represent and combine uncertain data sources.
Unlike classical methods that aim to produce a single “optimal” estimate, it allows reasoning in terms of degrees of belief rather than a single truth. Concretely, each information source (for example, a map derived from satellite imagery or an existing dataset) does not just provide a binary answer, but an associated confidence measure for different possible scenarios.
The method then allows these sources to be combined by distinguishing three elements: what is confirmed by multiple sources, what is in disagreement, and what remains uncertain due to insufficient information. One of the major advantages of this approach is that it does not force an immediate decision when there are contradictions between data.
On the contrary, it allows preserving and making visible areas of ambiguity, which is particularly useful when the sources are numerous, heterogeneous, or partially incompatible. Applied to Sumatra, this method improves the accuracy of the results while making visible the areas of divergence between the data.
For example, it highlights agricultural zones where coconut trees are confused with oil palms, while large industrial plantations are better identified and more consistently agreed upon among the different sources. Furthermore, our method provides more nuanced information.
This approach better represents uncertainty, often invisible but essential to understand the limitations of the data and to avoid an overly deterministic interpretation. Rather than binary maps (presence or absence of oil palm plantations), it becomes possible to offer gradual representations indicating different levels of confidence.
A high confidence threshold will result in smaller detected areas but with strong certainty, while a lower threshold will include more zones, at the cost of greater uncertainty. Users can then adjust this threshold according to their needs and produce maps consistent with their objectives, while remaining aware of the margins of uncertainty.
For example, in the context of monitoring oil palm plantations, an NGO may focus on areas where the confidence level is highest to document effectively present oil palm plantations, while an administration may include more uncertain areas to identify zones likely to be converted soon and direct field inspections.
From the cartographic revolution to “map-washing” Bringing together different actors and civil society in this way to produce more readable maps is a way to fight against what some researchers have called the “”map-washing””, or a process of disseminating spatial information of little value to users, but contributing to building or guiding a particular narrative.
If the social sciences researcher Rory Padfield and his colleagues haveconceptualized this driftto analyze cartographic tools used to promote an image of environmental transparency in the palm oil industry, this concept can be extended to situations where the multiplication of maps and data ultimately complicates their interpretation and use.
The problem is not only that maps can be misleading: it is also that they become difficult to use. This situation paves the way for a more diffuse form of instrumentalization, not through direct manipulation of the data, but through opportunistic selection of the map best suited to a given objective.
Thus, themap-washingdoes not rely solely on a deliberate communication strategy: it can also emerge from an excess of poorly articulated information. Producing more and more maps does not guarantee better knowledge.
The central issue then becomes shifting from a production logic to a usage logic: better defining the mapped objects, making uncertainties visible, and developing tools capable of articulating different sources of information.
Carl Bethuel received funding from the Brittany Region and the National Research Agency through the ANR Palmexpand project (ANR-20-CE03-0004) –ref.
Why the abundance of maps doesn’t necessarily help us be better informed –https://theconversation.com/why-the-profusion-of-maps-does-not-necessarily-help-us-be-better-informed-282287