Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Benoît Smets has a total of 210 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Unravelling information on impactful geo-hydrological hazard events with HazMiner, a multilingual text mining method developed through a global scale coverage application

    by Bram Valkenborg, Olivier Dewitte, I Wayan Gita Krishna Dananjaya, Benoît Smets

    The incidence and impacts of geo-hydrological hazards such as floods, flash floods and landslides are changing globally due to anthropogenic environmental changes and increased exposure driven by population growth. Reliable datasets on geo-hydrological hazards are essential to deepen our understanding of these hazards and their impacts. However, existing geo-hydrological hazard datasets contain data gaps leading to biased interpretations, especially in low- and lower-middle-income countries where populations are commonly the most impacted. Text mining offers new opportunities in documenting geo-hydrological hazards by automatically extracting information from large text corpora. Despite its potential, current methodologies are not adapted to improve documentation in data-scarce contexts. We introduce HazMiner, a paragraph-based text mining method designed to document the location, timing and impact of geo-hydrological hazards through transformer-based natural language processing models across multiple languages and at various scales. Applied here globally on 6 366 905 news articles published from 2017 through 2024 in 58 languages, HazMiner extracted 15 313 flood, 6836 landslide and 2258 flash flood events with a known location and time information and, in some cases, impact data. Compared to existing hazard datasets, HazMiner significantly improved hazard documentation, reducing the data gaps in many regions, especially in low- and lower-middle-income countries. This new versatile multilingual method and its dataset advance the accessibility and use of text mined natural hazard data.

  • Open Access

    Comment on egusphere-2026-722

    by Bram Valkenborg, Olivier Dewitte, Benoît Smets

    The incidence and impacts from geo-hydrological hazards (GH) such as floods, flash floods and landslides are changing globally due to anthropogenic environmental changes and increased exposure driven by population growth. Reliable datasets on GH are essential to deepen our understanding of these hazards and their impacts. However, existing GH datasets contain data gaps leading to biased interpretations, especially in the Global South where populations are commonly the most impacted. Text mining offers new opportunities in documenting GH by automatically extracting information from large text corpora. Despite its potential, current methodologies are not adapted to improve documentation in data-scared contexts. We present HazMiner, a paragraph-based text mining method designed to document the location, timing and impact of GH through large language models across multiple languages and at various scales. Applied here globally on 6,366,905 news articles published from 2017 through 2024 in 58 languages, HazMiner extracted 21,411 flood, 7,659 landslide and 3,606 flash flood events with known location and time information and, in some cases, impact data. Compared to existing hazard datasets, HazMiner significantly improved hazard documentation, reducing the data gaps in many regions, especially in the Global South. The new versatile multilingual method and its dataset advances both text mining and natural hazard research.

  • Open Access

    Unstable Slopes and Shifting Landscapes: Slow-moving landslides in the East African Rift

    by Antoine Dille, Matthias Vanmaercke, Toussaint Mugaruka Bibentyo, Floriane Provost, Benoît Smets, Olivier Dewitte

    Human activities are transforming tropical mountain landscapes at unprecedented rates through deforestation, agricultural expansion, and urbanization. These changes amplify the frequency and magnitude of geo-hydrological hazards such as landslides. While shallow, rapid landslides are well documented, the controls on the activity and dynamics of large, slow-moving landslides (SML) remain much less understood, despite their persistent impacts on communities and sediment dynamics.This study demonstrates how the combined use of radar and optical Earth observation data enables the detection, mapping, and monitoring of deep-seated landslides across vast and remote tropical regions such as the Albertine Rift. By mapping and comparing more than 120 active and 3,000 historical landslides distributed along the ~1,500 km Rift branch, we reveal how climatic, lithological, tectonic, and anthropogenic factors jointly control their occurrence.We further analyse multi-year landslide dynamics across contrasting environments, supported by unique ground-based validation datasets built on years of fieldwork in the region, and provide detailed insights into failure mechanisms of recent catastrophic landslides in the area. Altogether, this work delivers a unique regional-scale assessment of SML activity in tropical environments and highlights how landscape and human-driven land use changes can modulate their behaviour. It offers new perspectives on how environmental transformations shape landscape evolution, geo-hydrological hazards and sediment transfer in rapidly changing mountain regions.

  • Open Access

    Comment on egusphere-2026-722

    by Bram Valkenborg, Olivier Dewitte, Benoît Smets

    The incidence and impacts from geo-hydrological hazards (GH) such as floods, flash floods and landslides are changing globally due to anthropogenic environmental changes and increased exposure driven by population growth. Reliable datasets on GH are essential to deepen our understanding of these hazards and their impacts. However, existing GH datasets contain data gaps leading to biased interpretations, especially in the Global South where populations are commonly the most impacted. Text mining offers new opportunities in documenting GH by automatically extracting information from large text corpora. Despite its potential, current methodologies are not adapted to improve documentation in data-scared contexts. We present HazMiner, a paragraph-based text mining method designed to document the location, timing and impact of GH through large language models across multiple languages and at various scales. Applied here globally on 6,366,905 news articles published from 2017 through 2024 in 58 languages, HazMiner extracted 21,411 flood, 7,659 landslide and 3,606 flash flood events with known location and time information and, in some cases, impact data. Compared to existing hazard datasets, HazMiner significantly improved hazard documentation, reducing the data gaps in many regions, especially in the Global South. The new versatile multilingual method and its dataset advances both text mining and natural hazard research.

  • Open Access

    A new global dataset of geo-hydrological hazards and their impacts automatically extracted from online news articles

    by Bram Valkenborg, Olivier Dewitte, Benoît Smets

    Environmental change and rapid population growth are altering the impacts of floods, landslides and flash floods. The Global South is disproportionally affected by these changes, resulting into an uneven impact of these geo-hydrological hazards compared to the Global North. Comprehensive global documentation of geo-hydrological hazards is needed to improve our understanding of these hazards, yet this remains challenging. Existing data collection approaches—such as remote sensing, empirical news article screening; and field-based surveys—have limitations, constraining our ability to accurately analyze distribution, impacts and trends in geo-hydrological hazard occurrence. Moreover, most global datasets suffer from various geographical, linguistic and socio-economical biases.To further address these challenges, we introduce a new global dataset documenting geo-hydrological hazards automatically extracted from online news articles by a large language model-based text mining algorithm, called HazMiner. A total of 6 366 905 news articles published in 58 languages from 2017 until 2025 were analyzed. The resulting dataset includes the location, timing and impact of 21 411 floods, 7 659 landslides and 3 606 flash floods. Compared to EM-DAT, a well-established global disaster dataset, our dataset documents 31 150 more geo-hydrological hazard events over the same period. Among these, 784 events resulted in at least one but fewer than ten fatalities and therefore do not meet one of EM-DAT inclusion criteria, collectively accounting for 3,578 fatalities.Spatially, these impactful hazards occur in densely populated areas and with floods primarily located along rivers, and landslides and flash floods concentrated in mountainous regions. Temporally, floods and flash floods show seasonal trends for both hemispheres. Furthermore, 30 810 geo-hydrological hazard events do not report any fatalities, providing a broader interpretation of these hazards at the global scale compared to existing global disaster datasets. This dataset offers a new detailed global view of the hazards and has the potential to improve our understanding of their spatial-temporal occurrence and their associated impacts and risks.

  • Open Access

    Unravelling information on impactful geo-hydrological hazard events with HazMiner, a multilingual text mining method developed through a global scale coverage application

    by Bram Valkenborg, Olivier Dewitte, Benoît Smets

    The incidence and impacts from geo-hydrological hazards (GH) such as floods, flash floods and landslides are changing globally due to anthropogenic environmental changes and increased exposure driven by population growth. Reliable datasets on GH are essential to deepen our understanding of these hazards and their impacts. However, existing GH datasets contain data gaps leading to biased interpretations, especially in the Global South where populations are commonly the most impacted. Text mining offers new opportunities in documenting GH by automatically extracting information from large text corpora. Despite its potential, current methodologies are not adapted to improve documentation in data-scared contexts. We present HazMiner, a paragraph-based text mining method designed to document the location, timing and impact of GH through large language models across multiple languages and at various scales. Applied here globally on 6,366,905 news articles published from 2017 through 2024 in 58 languages, HazMiner extracted 21,411 flood, 7,659 landslide and 3,606 flash flood events with known location and time information and, in some cases, impact data. Compared to existing hazard datasets, HazMiner significantly improved hazard documentation, reducing the data gaps in many regions, especially in the Global South. The new versatile multilingual method and its dataset advances both text mining and natural hazard research.

  • Open Access

    A multilingual tool for the documentation of impactful geo-hydrological hazards using online news articles: a worldwide application

    by Bram Valkenborg, Olivier Dewitte, Benoît Smets

    Efforts worldwide aim to collect detailed information on the spatial and temporal distribution of natural hazards to improve our understanding of their occurrence and ultimately prevent their impacts. However, data on the location, timing, and impact of hazards remain scarce in many regions, even in the most exposed ones. Data collection methods are usually framed around earth observation approaches, sometimes combined with citizen science. Such approaches can be time-consuming, resource-intensive, and may fall short regarding data needs, especially at large scales. Combining these methods with complementary approaches could better address these challenges. We introduce a multilingual tool that uses natural language processing techniques to extract information on geo-hydrological hazards from online news articles. The tool is developed based on a worldwide application where we processed ~ 5.8 million articles published between 2017 and 2023 across 58 languages. The articles were extracted from GDELT (Global Database of Events, Language, and Tone), a global database monitoring events through online news articles. Using large language models, the tool analyzes articles at the paragraph level through three major steps: (1) filtering paragraphs for relevancy, (2) extracting information on the location (down to street level), timing, and impact, and (3) clustering information into events. This multilingual approach enabled the tool to extract and analyze 12.438 flood events, 1.312 landslide events, and 1.086 flash flood events globally for 2023 alone, providing ~ 20 times more data than current disaster databases and improving the coverage worldwide. In regions such as South and Central America, Europe, and Asia, where English is not the primary reporting language, non-English texts were the most important source of information. Especially in South and Central America, where non-English (primarily Spanish and Portuguese) paragraphs outnumbered English paragraphs by a factor of five. The proposed tool provides a new way to extract an unprecedented level of data on geo-hydrological hazards, forming a complementary source of information to existing methodologies. Beyond geo-hydrological hazards, the tool can be used to document other hazards, including earthquakes, wildfires, or volcanic activity. In addition, with this specific application, we provide a new extensive global dataset on impactful geo-hydrological hazards, which offers new opportunities for improving our understanding of these processes and their impact on continental to global scale.

  • Open Access

    Combining radar and optical satellite data to gather a comprehensive regional-scale dataset of the activity of slow-moving landslides in diverse tropical landscapes

    by Antoine Dille, Benoît Smets, Matthias Vanmaercke, Olivier Dewitte

    Slow-moving landslides (SML; mm year−1 to 100 m year−1) can be a ubiquitous geomorphic process in tropical mountain landscapes. Yet, answer to crucial questions such as what landscape characteristics exert the most important control on their spatial distribution (e.g., slope, connection to rivers, climate, lithology, tectonic setting, recent deforestation, degree of anthropogenic activity, etc.), or how does their dynamic behaviour responds to landscape changes (urbanisation, deforestation, etc.), remains elusive – and is typically relying on information collected on single or a few landslide(s). Intrinsically complex, obtaining large-scale datasets with dense surface displacement measurements is even more so in the tropics, where field access is typically difficult, and rapid vegetation changes and persistent cloud cover hamper the use of satellite remote sensing. In this work, we attempt to overcome these limitations by exploiting synergies between spaceborne sensors (i.e., radar and optical) and deformation measurement techniques (i.e., interferometry and sub-pixel image correlation), to obtain multi-year datasets of the activity of SML in the western branch of the East African Rift (wEAR). Characterised by a large natural landscape and climatic diversity, the wEAR is exemplative of many tropical mountain regions, i.e., i) affected by large-scale land use changes and ii) disproportionately high landslide impacts and iii) largely overlooked in landslide research. We collected a spatio-temporal inventory containing characterised by varying level of activity and behaviours, and located in contrasting environments. This regional-scale dataset will form the foundation for untangling the intricate influences of climate, lithology, tectonics and man-made environmental changes on the occurrence and activity of SML. By investigating their interaction with river system, we also aim at estimating how they contribute to controls on river sediment budgets, regional erosion rates, channel network evolution and flooding patterns – key for our understanding of landscape evolution, sediment budgets and geo-hydrological hazards. Overall, this work aims at moving forward our understanding of a key geomorphic process in severely under-researched types of environments subject to rapid changes. This is not only essential for a better hazard assessment, but also for comprehending how (human-induced and/or natural) environmental changes affect these landscapes and the sediment dynamics.

  • Open Access

    Raising awareness to geo-hydrological hazard risks in African countries: A guide booklet for stakeholders, policy markers and the public at large

    by Olivier Dewitte, Joseph Martial Akame, Diawara Bandiougou, Özlem Adiyaman Lopes, Antoine Dille, François Kervyn, Benoît Smets, Caroline Michellier, Camille François

    Many regions of Africa are exposed to a large variety of geo-hydrological hazards such as earthquakes, volcanic eruptions, landslides, floods, karst collapses and large urban gullies. Despite the soaring impacts on population, infrastructure and the environment associated with the occurrence of these hazard risks, most regions are under-studied. In addition to this lack of information, stakeholders, policy makers and the public at large remain relatively poorly aware of the hazard and risk problems, whether it is about their causes, their impact, and/or their mitigation. This overall lack of knowledge and awareness is associated with an aggravation of the impacts as the growing and vulnerable population of these regions, in search for new settlements and opportunities, is often moving towards areas that are more prone to natural hazards. This is in this context that UNESCO supports the preparation and dissemination of a guide booklet on geo-hydrological hazards for stakeholders, policy makers and the general public. The booklet targets ten African countries (Angola, Burundi, Cameroon, Central African Republic, Chad, Democratic Republic of the Congo, Equatorial Guinea, Gabon, Republic of the Congo, São Tomé and Príncipe) that are covered by the UNESCO regional office of Yaoundé. The aim of this work is to raise collective awareness of the need to prevent natural hazard risks at local, regional and national levels in order to ensure the protection of populations and promote the sustainable development of territories. In this way, UNESCO aims to guide and advise the ten African countries by providing them with useful and practical information.

  • Open Access

    Thermal remote sensing of lava lakes: a comparison of approaches

    by Sofie Rolain, Dries Peumans, Benoît Smets

    Satellite-based thermal remote sensing is a useful tool for monitoring volcanoes. It involves detecting thermal anomalies, called 'hotspots,' and calculating the radiative energy emitted by volcanic activity. Various volcanic hot spot detection algorithms already exist in the literature. However, every algorithm has its advantages and disadvantages, as they are limited depending on the tradeoffs made during algorithm development, the sensor used for aquisition, and the geometry of acquisition. Depending on the algorithm used, different results are obtained from the same data and, hence, different interpretations can be made in terms of, e.g., energy emitted, effusion rates, and eruption duration. In the present work, we aim at creating a new hotspot detection algorithm using MODIS and VIIRS imagery, which allows us to efficiently look at the dynamics of thermal emissions coming from persistent lava lakes, i.e., bassins of lava maintained molten through thermal convection and outgassing. We investigate the applicability of sensor fusion ideas, using multiple bands, and incorporating cloud cover information. We expect that by combining all available data the robustness of the detection process will increase.