166 research outputs found
Sort by
Harnessing EO and Natural Experiments for Urban Development: The UDENE Approach
Full text linkUrban Development Explorations using Natural Experiments (UDENE) is a forward-looking initiative under the Horizon Europe program that merges Earth Observation (EO) technologies with urban planning to tackle pressing urban challenges. By utilizing Copernicus satellite imagery and organizing local in-situ data into interoperable data cubes, UDENE provides a comprehensive framework for data-driven decision-making. Further, it is applied the concept of “natural experiments” - real-life changes analyzed with the rigor of controlled studies - to uncover causal relationships in urban development. A primary goal is to incorporate structured urban data into the broader Copernicus data cube federation, enabling consistent analysis of urban impacts across different times and locations. To support this, UDENE develops advanced sensitivity analysis methods for validating and applying multivariate causal models, enhancing predictions on factors such as air pollution, urban heat, mobility, and disaster resilience. To close the gap between high-level EO technologies and real-world planning needs, we are introducing the three core tools: the UDENE’s Data Cube, which populates in-situ data EO based analysis-ready data and datasets; the Exploration Tool, which empowers planners and policymakers to simulate, assess, and visualize urban interventions; and a matchmaking tool connecting users with EO-based services. Together, these tools foster informed urban strategies grounded in EO data and causal inference
Signal-Domain Guided Deep Learning for Gap-Filling of XCO and XCH4: A Masked Spatio-Temporal Fusion of TROPOMI and GEOS-Chem (2019–2023)
No full textLong-term, high-resolution monitoring of carbon monoxide (CO) and methane (CH4) is essential for understanding their spatiotemporal variability and guiding climate mitigation strategies. However, satellite observations like TROPOMI are often incomplete, and existing fusion methods have limitations in accuracy and continuity. This study proposes a signal-domain fusion approach combining 3D discrete cosine transform (DCT) and singular value decomposition (SVD) to integrate TROPOMI data with GEOS-Chem simulations. A lightweight residual U-Net is employed to refine the initial reconstruction by learning the residual field using meteorological drivers and model outputs, guided by a masked loss. The method produces global 0.25° and China-specific 0.05° daily gap-free XCO and XCH4 datasets from 2019 to 2023. The fused results outperform GEOS-Chem and are comparable or superior to TROPOMI, with R² values of 0.92 for XCO and 0.85 for XCH4. Trend analysis reveals regional patterns such as XCO increases in North America and declines in Eastern China, and widespread CH4 growth. High-resolution data captures enhancements during the 2022 Chongqing wildfires, with average increases of 17.1 ppb in XCO and 24.5 ppb in XCH4, and reveals lower XCH4 increases over rice-growing areas compared to TROPOMI, with overestimation reduced by 17–26 %, and stronger XCO reductions, with satellite underestimations up to 38 %. These results highlight agricultural contributions and policy impacts. This approach effectively reconstructs missing observations and enhances the utility of satellite–model data for atmospheric research and emission assessments. The generated daily gap-free datasets are publicly available at https://doi.org/10.5281/zenodo.17936461
A Stacking Ensemble Technique to Predict Signal Path Loss via 3D GIS
Full text linkThe recent developments in cellular communication technologies, especially the emergence of 6G, have increased the need for accurate and reliable signal path loss prediction models. The accuracy of predictions is reduced because traditional empirical approaches often fail to take into account the complex relationships between radio signals and the three-dimensional urban environment. Therefore, integrating advanced machine learning algorithms with diverse geographic data offers a promising direction for improving prediction performance and supporting next-generation network planning. This paper introduces an integrated methodology that combines Geographic Information Systems (GIS) with stacking ensemble machine learning models to enhance signal path loss prediction. The study made several key contributions, which are outlined below: (I) A GIS-based framework has been developed to integrate the Digital Twin (DT) of the study area with machine learning-based path loss models, incorporating 3D geographic data such as terrain height and building elevations. (II) The study assesses binary hybrid algorithms by examining three ensemble learning models (Gradient Boosting Machine (GBM), Extreme Gradient Boosting (XGBoost), and CatBoost). The fusion of 3D spatial data with ensemble learning algorithms has led to notable advancements in mobile network design, improving the accuracy of signal attenuation predictions. (III) Lastly, the paper emphasizes the potential of GIS-assisted machine learning techniques for future network deployments, including applications in DT, 6G, and beyond
Next-Generation Snow Cover Monitoring in Near Real Time: Evaluating the EUMETSAT H SAF H43 Product from MTG-FCI across the European Alps
Full text linkReliable snow cover mapping in mountainous regions is critical for hydrology, climate monitoring, and hazard risk management. This study presents the initial evaluation of the EUMETSAT H SAF H43 snow cover extent product—the first derived from the Flexible Combined Imager (FCI) aboard Meteosat Third Generation (MTG)—over the European Alps. The assessment was performed by comparing H43 with its predecessor H34, using daily MODIS MOD10A1 NDSI-based binary snow maps as the reference for the 2024–2025 winter season. Evaluation metrics, including Probability of Detection (POD), False Alarm Ratio (FAR), and Overall Accuracy (ACC), were calculated based on pixel-level agreement across December to February. To investigate terrain-dependent classification performance, the spatial distribution of false alarms (B) and missed detections (C) was analyzed across elevation zones and terrain aspect classes derived from the MODIS 1 km MODDEM product. Wind rose visualizations revealed that both products exhibit classification uncertainty in mid-elevation ranges (1000–2000 m), with H34 showing higher miss rates and H43 slightly more false alarms. In higher elevation bands (≥2000 m), H43 demonstrated improved stability across slope orientations and generally lower error rates. These findings highlight the enhanced snow retrieval capability of the H43 product in complex alpine environments and support its application in near-real-time snow monitoring
Version 8 IMK/IAA MIPAS measurements of ClO
Full text linkGlobal distributions of chlorine monoxide (ClO) were retrieved from infrared limb emission spectra recorded with the Michelson Interferometer for Passive Atmospheric Sounding (MIPAS), covering the time period from July 2002 to April 2012. The retrieval was performed by constrained non-linear least squares fitting using spectral lines in the fundamental band of ClO around 844 cm−1. The vertical resolution of V8 ClO is 4 km at 18–20 km and 7.5–9.5 km at 40 km altitude. The considerable improvement at 40 km with respect to the previous V5 data version is achieved by extension of the spectral range for retrieval of upper stratospheric ClO. Errors are by far dominated by measurement noise and increase from 0.4–0.5 ppbv at 20 km to 0.8 ppbv at 50 km altitude. Thus, in general, individual ClO profiles are noisy, and profile averaging has to be performed for, e.g., analysis of the upper stratospheric maximum. However, strongly enhanced lower stratospheric ClO amounts of more than 1.5 ppbv during polar winter are well detected in single measurements. Along with the standard representation of the data, an alternative coarse grid representation that obviates the need to apply averaging kernels in certain situations is also provided. Due to improved modeling of the atmospheric continuum and the instrumental offset, the high bias in upper stratospheric ClO that had particularly affected the previous V5 data over the period 2005–2012 has been removed. A comparison with ClO measurements of the Microwave Limb Sounder (MLS) on the Aura satellite shows good agreement between the lower stratospheric enhancements observed by the two instruments in polar winter. There is also good agreement between the upper stratospheric ClO amounts observed in the northern hemisphere and at southern hemispheric low latitudes. With the support of simulations from the Earth system model ECHAM/MESSy Atmospheric Chemistry (EMAC), deviations between the ClO amounts of MIPAS and MLS in the Antarctic lower stratosphere during July and in the upper stratosphere, especially at southern mid- and high latitudes during winter, are attributed to the different local solar times of the measurements.</p
Probabilistic hierarchical interpolation and interpretable neural network configurations for flood prediction
Full text linkThe past few years have witnessed the rise of neural networks (NNs) applications for hydrological time series modeling. By virtue of their capabilities, NN models can achieve unprecedented levels of performance when learning how to solve increasingly complex rainfall-runoff processes via data, making them pivotal for the development of computational hydrologic tasks such as flood predictions. The NN models should, to be considered practical, provide a probabilistic understanding of the model mechanisms and predictions and hints on what could perturb the model. In this paper, we developed two NN models, i.e., Neural Hierarchical Interpolation for Time Series Forecasting (N-HiTS) and Network-Based Expansion Analysis for Interpretable Time Series Forecasting (N-BEATS) with a probabilistic multi-quantile objective and benchmarked them with long short-term memory (LSTM) for flood prediction across two headwater streams in Georgia and North Carolina, USA. To generate a probabilistic prediction, a Multi-Quantile Loss was used to assess the 95th percentile prediction uncertainty (95 PPU) of multiple flooding events. Extensive experiments demonstrated the advantages of hierarchical interpolation and interpretable architecture, where both N-HiTS and N-BEATS provided an average accuracy improvement of ∼ 5 % over the LSTM benchmarking model. On a variety of flooding events, both N-HiTS and N-BEATS demonstrated significant performance improvements over the LSTM benchmark and showcased their probabilistic predictions by specifying a likelihood objective.</p
Unique microphysical properties of small boundary layer ice particles under pristine conditions on Dome C, Antarctica
Full text linkThe Antarctic plateau, one of the coldest and cleanest regions of our planet, experiences almost exclusively frozen precipitation. Understanding the microphysical properties of inland Antarctic boundary layer ice particles with sizes below a few hundred micrometers is essential to improve atmospheric models and accurately validate remote sensing data for this region. Currently, only a small number of in situ atmospheric measurements exist for particle sizes smaller than 100 µm on the Antarctic plateau, performed over short measurement times. We present the first multi-week study of optical in situ measurements of boundary layer ice particle size, shape and morphological complexity for sizes down to 11 µm with a temporal resolution in the order of minutes, including a multi-day ice fog event. Classifying ice fog events with a lidar system, we found mean particle sizes smaller than 11 µm for ice fog events and of about 70 µm for cirrus precipitation and diamond dust events. The mean particle concentration of the ice fog at Dome C (3.9 L−1) is found to be lower than in parametrisations of Arctic ice fog and lower than the concentration of anthropogenically influenced urban ice fog measured at Fairbanks, Alaska during a three-year study with the same instrument (90 L−1). Moreover, ice fog particles at Dome C are found to be more pristine than at Fairbanks. Our findings show that Antarctic boundary layer ice particles may need to be parametrised differently than their Arctic counterparts due to distinct conditions on the Antarctic plateau.</p
One-day repeat pass interferometry highlights the role of temporal baseline on digital elevation models retrieved from Sentinel-1
Full text linkDigital Elevation Models (DEMs) derived from Synthetic Aperture Radar (SAR) interferometry are a key data source for numerous geospatial applications, from hydrological modelling to environmental monitoring. The launch of Sentinel-1C in late 2025 introduces a new sensor into the Sentinel-1 constellation. This study evaluates the vertical accuracy of DEMs generated from interferometric image pairs acquired during the satellite's calibration phase. The analysis uses a set of image pairs with temporal baselines of 1, 6, and 12 d, over a test site in Angola, validated against ICESat-2 elevation measurements. The workflow includes interferometric processing, coherence assessment, and statistical error evaluation. Results indicate high accuracy for the 1 d pair (RMSE≈14.7 m) and moderate degradation for the 6 d pair (RMSE≈16.4 m), but a pronounced loss of accuracy for the 12 d pair (RMSE≈49.4 m), primarily linked to coherence loss in vegetated areas. Coherence and elevation error distributions reveal clear land cover and slope dependencies, with lower performance in forested and steep terrain. These findings should be regarded as indicative due to the limited number of suitable image pairs for the calibration phase. However, this early assessment provides an important reference point for future Sentinel-1A/C DEM generation studies, informing both methodological refinement and application planning in SAR-based topographic mapping.</p
A Comparative Study of Nature-Inspired Optimization Techniques for MLP-Based Sentinel-2 Image Segmentation
Full text linkAccurate land cover classification using Sentinel-2 satellite imagery remains a critical challenge in remote sensing due to spectral complexity and spatial heterogeneity. This study presents a comprehensive evaluation of Multi-Layer Perceptron (MLP) models optimized with nature-inspired algorithms for Sentinel-2 image segmentation. We compare five optimization approaches Ant Colony Optimization (ACO), Whale Optimization Algorithm (WOA), Particle Swarm Optimization (PSO), Genetic Algorithm (GA), and Artificial Bee Colony (ABC) to enhance MLP performance for classifying five key land cover types: urban areas, agricultural fields, sparse vegetation, water bodies, and forests. Our optimized MLP architecture achieves superior performance with 90.8% overall accuracy, 90.7% F1-score, 0.883 Cohen’s Kappa, and 0.981 ROC-AUC, representing a 7.2% improvement over the best-performing nature-inspired algorithm (GA/WOA at 83.6% accuracy). Class-specific analysis reveals high accuracy for water bodies (94.2% F1-score) and forests (91.6%), while urban areas (87.4%) and sparse vegetation (82.7%) present greater challenges due to spectral similarities. The study demonstrates that hybrid optimization, combining algorithmic tuning with expert refinement, yields the most robust results for operational land cover mapping. Key findings highlight GA’s effectiveness in handling class imbalance and WOA’s strength in rare class detection. Computational efficiency (2–4 hours training time) further supports the model’s feasibility for large-scale applications. This research advances Sentinel-2 segmentation methodologies while providing practical insights for environmental monitoring, precision agriculture, and urban planning
Limited atmospheric iron availability increase during the Pleistocene-Holocene transition in the Northern Hemisphere
Full text linkIron (Fe) availability modulates phytoplankton blooms in High-Nutrient Low-Chlorophyll (HNLC) regions, i.e., ocean areas characterized by an abundance of major nutrients but low marine productivity. Fe can be delivered to the oceans through atmospheric dust deposition, making ice cores unique archives for reconstructing past changes in aeolian Fe deposition. However, while it is known that during dustier periods atmospheric Fe depositions increased, uncertainties remain regarding the fraction of Fe actually available to phytoplankton. Here, we present evidence from the EGRIP ice core (Greenland), which allows insights into atmospheric aerosol deposition over the Fe-limited North Pacific Ocean, during the Pleistocene-Holocene transition (10.3–13.0 ka). Results show that, in contrast to the 17-fold enhancement in total Fe concentration, dissolved Fe increased only modestly (+29 %) during the Younger Dryas compared to the Early Holocene, likely due to prevailing alkaline aerosol conditions reducing its solubility. This finding supports the hypothesis that factors other than atmospheric Fe deposition (e.g., stronger water stratification, sea-ice extent, volcanic eruptions, iron remobilization from sediments), play a more relevant role in regulating marine net primary productivity in the HNLC North Pacific Ocean over the last glacial transition