Deliverables
Past and present marine citizen science around the globe: A cumulative inventory of initiatives and data produced
The article examines 1,267 marine citizen science initiatives, highlighting their growth, geographical distribution, and focus areas. It underscores the role of citizen participation in addressing marine environmental challenges while identifying gaps in data accessibility and standardization. The study emphasizes the need for better data management practices to enhance the impact of citizen science on marine research and conservation.
Daily high-resolution surface PM2.5 estimation over Europe by ML-based downscaling of the CAMS regional forecast
The article presents the S-MESH (Satellite and ML-based Estimation of Surface air quality at High resolution) framework for estimating daily PM2.5 concentrations at 1 km resolution across Europe. Using a stacked XGBoost model, the approach integrates satellite observations, meteorological data, and the CAMS regional forecast to improve air quality assessments. The results demonstrate higher accuracy and better detection of pollution events compared to existing methods, with significant implications for air quality monitoring.
Data fusion of sparse, heterogeneous, and mobile sensor devices using adaptive distance attention
The article presents a novel method for integrating diverse environmental sensor data to improve spatial predictions. The authors propose an adaptive distance attention framework combining geostatistical techniques like kriging with deep learning models to enhance data fusion. Applied to case studies involving topography and air pollution, the method demonstrates improved predictive accuracy over traditional approaches, offering a scalable solution for environmental monitoring in complex and data-sparse regions.
Estimating surface NO₂ concentrations over Europe using Sentinel-5P TROPOMI observations and Machine Learning
The article presents a novel approach to estimating daily surface NO₂ concentrations at 1 km resolution across Europe. Using the S-MESH framework, the study employs an XGBoost model that integrates TROPOMI satellite observations with auxiliary data like night light radiance and meteorological factors. The model achieves robust performance, highlighting the value of satellite data and machine learning in air quality monitoring. It also explores feature importance using explainable AI techniques like SHAP.
Opening science to society: how to progress societal engagement into (open) science policies
The article explores the complex interplay between species traits, environmental factors, and ecological interactions shaping biodiversity patterns. Through a robust analytical framework, it evaluates how functional diversity influences ecological stability and adaptation in dynamic ecosystems. This study provides insights into preserving biodiversity amid global environmental changes. For further details, you can access the article.
