Fant 2799 publikasjoner. Viser side 1 av 280:
2026
2026
ML-based data fusion of model, satellite, and ground observations for 1-km PM2.5 mapping over Europe
2026
Bioaerosols interact with society and environment in a multi-faceted way. Information about biological aerosols in the atmosphere is at high demand for medical practitioners and allergy sufferers, climate change researchers, agriculture and forestry industries, air quality forecasters, a variety of information added-value businesses, and many other stakeholders. However, the monitoring practices established over 70 years ago and barely changed since then are country-specific, with varying data availability and usage policy. These roadblocks slow down cross-disciplinary research and development of measures to understand and, upon necessity, control societal and environmental impacts of bioaerosols.A series of technological breakthroughs during last 10 years introduced a variety of automatic particle counters capable of bioaerosol monitoring in real time. They paved the way to the volunteering consolidation of European aerobiologists to establish the EUMETNET AutoPollen Programme (www.autopollen.net), laid down the foundation for the bioaerosol monitoring infrastructure with the EU Horizon SYLVA project (A SYstem for reaL-time obserVation of Aeroallergens, https://sylva.bioaerosol.eu), initiated developments of European standards and guidelines for the automatic bioaerosol measurements with the EURAMET project BioAirMet, and started the European standardization effort with CEN WG 39.The new technologies allow to observe bioaerosol concentration in real time, analyze vertical concentration profiles via remote-sensing, perform metagenomic analysis of bioaerosols with the 3rd generation DNA sequencing technique, and combine these observations with atmospheric composition models. Newly established regional networks have been connected to regional atmospheric composition models, which assimilate the real-time regional data to improve the forecasts. It changes the existing paradigm of bioaerosol observations as the new monitoring networks involve large-scale data handling infrastructure, which also includes numerical models as an interface between the different technologies and a bridge to users of information.The new observations heavily rely on sophisticated technologies, such as high-resolution image analysis, holography, multi-band scatterometry and fluorescence spectrometry, lidar-based remote sensing, and nanotechnology for DNA sequencing. A particle recognition task, the key challenge for the new devices, is solved via machine learning approaches. Technological complexity of the new instruments and large amounts of raw data they produce have been recognized, and a European-scale solution has been proposed by AutoPollen/SYLVA. AutoPollen is being converted into a EUMETNET operational programme with the SYLVA infrastructure as its technological backbone. The programme, with support of Copernicus Atmosphere Monitoring Service (https://atmosphere.copernicus.eu), ACTRIS aerosol monitoring network, and other stakeholders, will become operational from 2027. The central processing system will be hosted by Finnish Meteorological Institute with support of MeteoSwiss, Technical University of Munich, and all SYLVA partners. The pre-operational work of AutoPollen/SYLVA started already in 2025, owing to the efforts of the SYLVA consortium, its sister projects and collaborators. The programme is open for all European (and from outside Europe) groups performing automatic bioaerosol monitoring. AutoPollen offers technological and organizational support, community-developed bioaerosol monitoring solutions, and a motivated team of experts advancing the relevant research and applications.
2026
Low-cost sensor (LCS) networks can complement sparse regulatory monitoring, but their value depends on robust integration strategies that preserve data quality while exploiting dense spatial sampling. Here we assess the added value of incorporating validated LCS PM2.5 observations into the S-MESH (Satellite and ML-based Estimation of Surface air quality at High resolution) machine learning framework (Shetty et al., 2024, 2025) to generate continental-scale, 1 km resolution surface PM2.5 fields across Central Europe. Two integration strategies are evaluated for 2021–2022 within a stacked XGBoost architecture driven by satellite aerosol optical depth, meteorological predictors, and CAMS regional forecasts: a) using LCS data as an additional training target (LCST), and b) using LCS information as a model input feature (LCSI) via an inverse-distance-weighted spatial convolution layer that encodes local sensor influence. Relative to a baseline trained only on official monitoring stations, LCSI yields consistent performance gains, with RMSE reductions of ~15–20% in urban areas, whereas LCST provides less consistent improvement. The resulting high-resolution mapping product achieves skill comparable to the CAMS regional reanalysis, often considered as a modelling “gold standard” for European air-quality assessment, and in some evaluations surpasses it, with lower annual mean absolute error (2.68 vs 3.32 µg m⁻³) (Shetty et al., 2026). This demonstrates that a data-fusion ML approach including LCS information can deliver reanalysis-level performance at 1 km resolution while requiring only modest computational resources compared with running full chemical transport model reanalyses, enabling rapid updates and scalable deployment. SHAP-based attribution further suggests that LCSI improves the model’s ability to capture localized pollution variability, while performance degrades where sensor density is low, limiting representation of inter-urban transport.Although demonstrated in Europe, the underlying methodology, namely combining globally available satellite products and meteorology with quality-controlled LCS networks in a computationally efficient ML framework, has potential to strengthen air-quality assessment also in resource-limited settings where regulatory infrastructure is scarce. A requirement for this is that appropriate sensor calibration/validation workflows are in place and equitable partnerships support sustainable sensor deployment and data stewardship. Shetty, S., Schneider, P., Stebel, K., Hamer, P. D., Kylling, A., and Koren Berntsen, T.: Estimating surface NO2 concentrations over Europe using Sentinel-5P TROPOMI observations and Machine Learning, Remote Sens. Environ., 312, 114321, https://doi.org/10.1016/j.rse.2024.114321, 2024.Shetty, S., Hamer, P. D., Stebel, K., Kylling, A., Hassani, A., Berntsen, T. K., and Schneider, P.: Daily high-resolution surface PM2.5 estimation over Europe by ML-based downscaling of the CAMS regional forecast, Environ. Res., 264, 120363, https://doi.org/10.1016/j.envres.2024.120363, 2025.Shetty, S., Hassani, A., Hamer, P. D., Stebel, K., Salamalikis, V., Berntsen, T. K., Castell, N., and Schneider, P.: Evaluating the role of low-cost sensors in machine learning based European PM2.5 monitoring, Environ. Res., 291, 123558, https://doi.org/10.1016/j.envres.2025.123558, 2026.
2026
The growing adoption of low-cost sensors (LCSs) has significantly enhanced environmental monitoring by enabling widespread, community-driven data collection, particularly in regions requiring dense monitoring, and in regions with limited or no reference instrumentation. Increased public awareness and demand for dense environmental monitoring have resulted in extensive air quality and meteorological datasets from diverse sources. However, the integration of such datasets into regulatory frameworks and large-scale environmental monitoring remains challenging due to persistent issues related to data quality, standardization, and interoperability. To address these challenges, the FILTER (Framework for Improving Low-cost Technology Effectiveness and Reliability) approach developed by Hassani et al. (2025) provides a suite of algorithms to harmonize, quality-check, flag, and perform in-situ corrections on crowdsourced PM2.5 LCS datasets. While FILTER was initially designed and validated for static PM2.5 sensors, it has since been extended to address data quality challenges associated with the dynamics of mobile and wearable sensing. Across both static and mobile LCS platforms, FILTER employs a unified processing pipeline that generates measurement-level quality flags based on multiple statistical tests, to quantify the reliability of each observation. Quality control (QC) includes statistical tests to: (a) assess physical measurement consistency (range validity test), (b) detect flatline behavior (constant value test), and (c) identify abnormal patterns (spatiotemporal outlier detection test) using both historical trends and spatial comparisons with neighboring LCSs. Beyond these mandatory QC steps, more advanced statistical procedures incorporate relative (spatial correlation test) and absolute (spatial similarity test) comparisons with nearby LCSs, higher-quality instruments, and reference monitoring stations. For mobile and wearable sensing, FILTER has been specifically adapted to support pairwise comparisons between mobile sensors and comparisons with higher-accuracy nodes, accounting for operation under dynamic environmental and operational conditions. In this context, statistical comparisons are evaluated during rendezvous events, that is, periods in which the mobile sensor and a higher-accuracy node provide temporally coincident measurements. The modified framework retains the core principles of transparency, scalability, and sensor independence, while introducing additional steps to address motion-related artifacts, intermittent time series, and location-specific uncertainties. FILTER is developed in the open-source R environment. Its modular and hierarchical design allows flexible adaptation of quality control and correction workflows based on data availability, the spatiotemporal characteristics of LCS networks, and application-specific requirements. By improving data reliability and usability, FILTER enables crowdsourced LCS datasets to serve as a reliable complement to official monitoring networks for air quality management, urban- and regional-scale modeling, and policymaking. References Hassani, A., Salamalikis, V., Schneider, P., Stebel, K., and Castell, N.: A scalable framework for harmonizing, standardization, and correcting crowd-sourced low-cost sensor PM2. 5 data across Europe, J. Environ. Manage., 380, 125100, 2025.
2026
Urban nature-based solutions (NBS) are increasingly deployed to restore ecosystems, regulate microclimates, support biodiversity, and enhance wellbeing. Yet many remain short-lived: once installation and early monitoring end, maintenance budgets shrink, responsibilities become unclear, and socio–ecological performance declines. The EU BiodivNBS NatureScape project addresses this overlooked post-implementation phase by examining how NBS are cared for, governed, and experienced over time in seven European cities – Oslo, Dublin, Riga, Milan, Lisbon, Lublin, and St. Gallen.To strengthen long-term sustainability, NatureScape establishes Transformation Labs (T-Labs) at demonstration sites, including rain gardens in Lublin; community gardens in Oslo, Riga, Milan, and St. Gallen; school gardens in Lisbon; and goat-grazing vegetation management in Dublin. These T-Labs function as practice-based innovation spaces where municipal authorities, researchers, and community groups jointly observe socio–ecological dynamics, identify stewardship challenges, and co-develop adaptive responses. The approach extends conventional living labs by focusing on long-term socio–ecological change and governance arrangements that support NBS persistence.NatureScape integrates baseline assessments across five forms of capital (natural, social, human, manufactured, financial) with participatory workshops, PPGIS, citizen science, and systems tools such as causal loop diagrams and multi-criteria assessments. This mixed-methods design enables analysis of NBS as dynamic systems shaped by interactions between ecological conditions, institutions, and community practices.Early findings from Oslo, Riga, Lublin and St. Gallen reveal recurrent barriers: unclear responsibilities after project funding ends, limited resources for routine care and climate adaptation, insecure land tenure, weak alignment with municipal strategies, and uneven community participation. In St. Gallen, expectations to expand activities, actors, or spatial scope further increase complexity and demand stronger management capacities.This study presents the NatureScape framework for post-implementation NBS governance and demonstrates how T-Labs can: (i) shift perceptions of NBS from temporary projects to living infrastructures requiring continuous care; (ii) clarify and redistribute responsibilities and resources for long-term stewardship; and (iii) provide structured settings where new forms of cooperation and valuation can be tested and embedded in policy. Embedding co-maintenance and co-stewardship as core practices can help cities move beyond pilot projects toward durable, multifunctional NBS aligned with EU and global biodiversity frameworks and targets.
2026
This study presents insights from the EU Biodiversa+ NatureScape project (2025–2028). The project offers a new perspective for understanding nature-based solutions (NBS) in cities by focusing on the post-implementation phase, in which environmental justice in urban planning is put to the test.In recent years, cities have increasingly pursued NBS in urban development projects such as community gardens, green roofs, and temporary green spaces to support biodiversity while simultaneously improving human well-being. Despite growing recognition of NBS in urban planning, their potential for cities' socio-ecological transformation remains constrained by overlooked post-implementation challenges. While the planning and implementation of NBS already receive considerable attention, critical dimensions of environmental justice – distributive equity, accessibility, and procedural justice for continuous public participation and stakeholder engagement – become apparent only in the post-implementation phase. This phase is characterized by dynamic interactions between social and ecological components, shaping whether NBS are consolidated and sustained in ways that contribute in the long term to transformative effects and environmental justice, or whether they instead undermine these aims.NatureScape addresses this critical transition and its challenges in urban planning. Through transformation laboratories (T-Labs) in seven cities (Oslo, Dublin, Riga, Milan, Lisbon, Lublin, and St. Gallen), the research team explores two central questions: (1) What enablers and barriers in urban planning shape the post-implementation stewardship of urban NBS? (2) What governance mechanisms, strategies, and measures lead to the successful integration of urban NBS into urban planning to unfold their transformative potential for biodiversity-positive transitions and environmental justice?Initial findings from the T-Labs reveal crucial barriers. The post-implementation phase is often reduced to technical maintenance. Insufficient incorporation of NBS into urban planning is associated with fragmented institutions and responsibilities, weak strategic and instrumental anchoring, financial insecurity, and the erosion of institutional and political support.The project identifies interconnected governance mechanisms that could successfully integrate NBS into urban planning: adaptive planning processes, institutional anchoring that fosters shared ownership among stakeholders, co-management approaches with formal agreements, public planning frameworks, and institutional structures that support integrated action. Together, these mechanisms highlight stewardship as a pivotal principle for achieving just and biodiversity-positive urban futures.
2026
2026