Correction of high-frequency spectral losses is a major technical challenge of the eddy covariance (EC) technique. If not properly accounted for during post-processing, these losses can result in a systematic underestimation of the measured gas fluxes exchanged between the ecosystem and the atmosphere. To address this issue, several methods have been developed, with experimental approaches relying on the definition of a transfer function and its associated cut-off frequency to describe the EC system as a first order low-pass filter.

One still debated yet fundamental choice is whether to use power spectra or co-spectra to derive the system cut-off frequency. In this study, we present a systematic, multi-site, data-driven comparison of the these two methods. To do so, we used one year of CO₂ and H₂O data from all of 38 ICOS Class 1 and Class 2 stations (Integrated Carbon Observation System, www.icos-cp.eu), all equipped with a standard setup comprising the LI-7200 enclosed path analyser and the HS-50 sonic anemometer.

We showed that the corrections were limited for both approaches, especially for CO₂, ranging from 1 to 1.2, generally higher for H₂O, ranging from 1 to 2, and overall consistent across sites. This highlighted the good spectral performance of the enclosed path analyser as well as the effectiveness of the setup standardisation. Nonetheless, the results showed that differences in correction factors between the methods existed. They were analysed for all sites, separately for stable and unstable conditions. They increased with atmospheric stability and attenuation level, and decreased with measurement height above the canopy. In particularly, they were systematically the highest in stable conditions. However, when assessing the impact of the two corrections on cumulative u*-filtered fluxes, we found that rejections of most stable conditions through this standard post-processing filtering led to differences under 3% for CO₂ in 89% of sites and under 6% for H₂O in 79% of sites.

With this specific experimental setup, we suggest prioritising the co-spectral for two main reasons. First, sensor separation is a dominant part of the high-frequency attenuation and is treated experimentally in the co-spectral method, whereas the spectral approach relies on a fully theoretical formulation. Second, the spectral method requires a robust denoising procedure, which is not needed in the co-spectral approach. Finally, while recognising its crucial importance at a network level, we highlight the complexity of having a fully automatic pipeline for spectral corrections.

How to cite: Faurès, A., Papale, D., Nicolini, G., Sabbatini, S., and Heinesch, B.: A multi-site comparison of spectral and co-spectral approaches for correction of turbulent gas fluxes with ICOS set-up, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-21667, https://doi.org/10.5194/egusphere-egu26-21667, 2026.

Nitrogen oxides (N₂O, NO, and NO₂) serve as critical linkages connecting climate systems, ecosystems, and atmospheric chemistry, with soils acting as a primary natural source. Adopting a multi-scale framework spanning global, regional, and field scales, we systematically examine the spatiotemporal heterogeneity of nitrogen oxide emissions from cropland soils. Spatially, emissions exhibit a latitudinal gradient, decreasing from low to high latitudes, with hotspots concentrated in agriculturally intensive regions. Temporally, emissions display multi-scale rhythmic patterns aligned with crop growth stages, seasonal cycles, and diurnal variations, tightly coupled to soil carbon-nitrogen transformation processes. From the perspective of carbon-nitrogen coupling mechanisms, we reveal how land management practices—including nitrogen fertilization, conservation tillage, and precision irrigation—regulate emissions by modulating soil organic carbon content, carbon-nitrogen ratios, and pore structure. Concurrently, climate change drivers such as rising temperatures, elevated CO₂ concentrations, and extreme precipitation alter microbial-mediated carbon-nitrogen transformation efficiency, collectively shaping the core mechanisms governing nitrogen oxide emissions. A meta-analysis further investigates light effects on soil nitrogen oxide emissions, demonstrating significant impacts: light exposure increased N₂O and NO fluxes by 57.28% and 116.19%, respectively. Notably, heightened UV-B radiation reduced N₂O emissions by 6.85%, whereas shading increased them by 77.23%, with crop-specific responses observed. Mechanistically, light regulates emissions by modifying soil physicochemical properties and restructuring nitrogen-cycling microbial communities. Current emission mitigation faces challenges, including underdeveloped monitoring systems, limited prediction accuracy due to multifactor coupling complexities, and poor regional adaptability of existing technologies. Integrating multi-source data (field observations, remote sensing inversion, laboratory experiments) with advanced modeling approaches—such as climate-soil-crop coupling models and machine learning algorithms—offers viable pathways to enhance emission prediction precision and optimize mitigation strategies. Looking ahead, priorities include establishing multi-scale automated monitoring networks, developing carbon-nitrogen coupling-driven predictive models, promoting regionally tailored carbon sequestration and nitrogen emission reduction technologies, and combining policy incentives with public engagement to reduce uncertainties in global carbon-nitrogen cycle projections. These efforts aim to strengthen scientific support for sustainable agricultural development.

How to cite: Li, D. and Shen, W.: Nitrogen oxide emissions from cropland soil: spatiotemporal heterogeneity, carbon-nitrogen coupling mechanisms, and mitigation strategies, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-9022, https://doi.org/10.5194/egusphere-egu26-9022, 2026.

Machine Learning models are rapidly becoming popular for complementing, enhancing, and in some cases, replacing traditional numerical models. This study presents a data-driven framework for predicting 24-hour tropical cyclone intensification over the North Indian Ocean using supervised machine learning and ERA5 reanalysis data. Cyclones that formed over Bay of Bengal and the Arabian Sea during the period 1990–2024 are considered here.  We have integrated environmental parameters from ERA5 with intensity records from the IBTrACS archive, excluding early developmental stages and retaining only dynamically mature systems. Intensification is formulated as a binary classification problem based on the sign of the 24-hour change in maximum sustained wind speed. While this captures general strengthening behaviour, it does not distinguish between moderate and rapid intensification, nor does it estimate the magnitude of intensity change. Five machine learning models—Logistic Regression, Random Forest, Extra Trees, Support Vector Machine, and Multilayer Perceptron—are trained and evaluated. Results indicate that the Random Forest classifier has achieved the highest accuracy. Feature-importance analysis reveals strong physical consistency, highlighting the dominant roles of upper-level circulation, sea surface temperature, vertical wind shear, and atmospheric moisture in regulating short-term intensification. Cyclone Montha (2025) is used as a test case to illustrate the model's real-world applicability and is validated outside of historical data. The model-predicted intensification probability is estimated as 0.943, which indicates good performance. Although a single case study does not constitute statistical validation, this illustrates the applicability of data-driven models in tropical cyclone intensity estimation. The results encourage further investigations into the use of such data-driven models in tropical cyclone intensity prediction, which aids disaster management efforts.

How to cite: Madhu, D., Binu, N. M., and Ramesh, M. V.: Machine Learning-Based Prediction of Tropical Cyclone Intensification Over the North Indian Ocean Using ERA5 Reanalysis , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-20844, https://doi.org/10.5194/egusphere-egu26-20844, 2026.

Dense fog events across India severely disrupt aviation, surface transportation, and daily activities during winter months, with northern districts experiencing extended periods of visibility issues. Building upon the WRF-based ensemble fog forecasting over the Indo-Gangetic Plain and BOFFIN-Melbourne's Bayesian Decision Network framework, this study proposes a continuous risk monitoring and decision support system at district-level (admin-2).

The operational system will conduct daily continuous risk assessment, leveraging satellite observations from MODIS/VIIRS/INSAT-3D and the ECMWF IFS ensemble forecasts (51 members, 0.25° resolution) including probabilistic meteorological predictions of temperature, dewpoint, wind speed, boundary layer height, relative humidity profiles, and cloud cover, to characterize antecedent fog conditions and to establish baseline occurrence patterns.

 A Bayesian Network will integrate these layers to provide real-time short-term forecasts using pre-defined conditional probability tables which encode relationships between stable boundary layer conditions, radiative cooling, and regional fog formation mechanisms. The operational output of the algorithms will be in the form of traffic light decision matrix for each district: Green (Minimal/Low risk - Monitor), Yellow (Moderate risk - Be Aware), Orange (High risk - Be Prepared), Red (Extreme risk - Take Action).

This paper will present the validation results from pilot districts and the development framework for scaling to nationwide continuous risk assessment, demonstrating the system's potential for proactive decision-making in transportation management, aviation operations, and public safety advisories.

References

• Parde, Avinash N., et al. "Operational probabilistic fog prediction based on ensemble forecast system: A decision support system for fog." Atmosphere 13.10 (2022): 1608.
• Boneh, Tal, et al. "Fog forecasting for Melbourne Airport using a Bayesian decision network." Weather and Forecasting 30.5 (2015): 1218-1233.

How to cite: Guttikunda, S. K., Kalladath, N., Tucci, R. R., Ouma, J., Amdihun, A., and Dammalapati, S. K.: Fog Risk Monitoring and Assessment for India Using Bayesian Networks and ECMWF IFS Ensemble Prediction System, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-9696, https://doi.org/10.5194/egusphere-egu26-9696, 2026.

Understanding vegetation responses to climate variability is essential for assessing long-term ecosystem dynamics. Leaf Area Index (LAI) is a widely used variable to characterise vegetation state and productivity. However, attributing observed global LAI trends to specific climatic drivers remains challenging due to non-linear interactions, strong spatial heterogeneity, and scale-dependent processes.

This study is conducted within the framework of the PROFECIA project, which aims to improve the monitoring and interpretation of vegetation responses to climate change by combining remote sensing observations and artificial intelligence techniques. We analyse global LAI trends over the period 1982–2022 using the GEOV2-AVHRR long-term satellite record and examine their relationship with trends in key climatic variables obtained from the ERA5 reanalysis, including temperature, precipitation, radiation, and several indicators of water availability and drought conditions. All trends are computed consistently over the 1982-2022 temporal record to ensure a homogeneous assessment of long-term vegetation–climate relationships at the global scale.

The vegetation–climate relationships are modelled using a suite of machine learning approaches, including tree-based methods and neural networks, designed to capture non-linear responses across diverse climatic and ecological conditions. Particular emphasis is placed on the role of the training strategy: different spatio-temporal sampling schemes are evaluated to assess their impact on model performance, robustness, and generalisation capability when analysing long-term trends at the global scale.

To move beyond purely predictive modelling, the study systematically applies explainable artificial intelligence (XAI) techniques to interpret the trained models. Methods such as SHAP-based attribution and partial dependence analyses are used to quantify the relative contribution of individual climatic drivers to observed LAI trends and to examine how these contributions vary across regions and time periods.

Overall, this work highlights the importance of combining robust machine learning training strategies with interpretability tools to improve the attribution of long-term vegetation trends to climatic drivers, providing new insights into global vegetation–climate interactions over the last four decades.

How to cite: García-Diaz, D., Aguilar, F., Schauman, S., and Verger, A.: Machine learning analysis of global LAI trends and their relationship with climate variability (1982–2022), EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-10657, https://doi.org/10.5194/egusphere-egu26-10657, 2026.

Particulate matter (PM) pollution represents a significant public health concern, in Lima, Peru. This issue is further compound by the lack of accurate forecasting tools due to limited monitoring networks. This study addresses this gap by developing and validating a hybrid model combining a Multilayer Perceptron (MLP) neural network and a type of rule-based Cellular Automata (CA) simulation. This model simulates and forecasts the spatiotemporal dispersion of PM10 and PM2.5. Using a decade of historical PM data (2015-2024) from seven monitoring stations and NASA's meteorological data, an optimized MLP was trained to learn the complex, non-linear transition rules from 47 engineered features. The model demonstrated remarkable performance in historical validation (R² > 0.90), outperforming standard baseline models. When fed with weather forecast data, the model can operate as an Early Warning System (EWS), providing a reliable prediction horizon to anticipate the exceedance of Air Quality Standards. The resulting hotspot maps accurately identify high-risk areas, confirming the potential of this hybrid model as a robust, proactive, and quantitative tool for air quality management and public health protection in complex urban environments.

How to cite: Maita, B., Condezo, P., Llanto, J., Huaman, S., and Sanabria, J.: A Hybrid Neural Network and Cellular Automata Model for spatiotemporal Forecasting of PM10 and PM2.5 in Lima, Peru, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-10823, https://doi.org/10.5194/egusphere-egu26-10823, 2026.

Fine particulate matter (PM_2.5) is among the foremost environmental determinants of human health, contributing to cardiovascular disease, respiratory illness, and premature mortality. In rapidly urbanizing regions of the Global South, accurate spatial characterization of PM_2.5 exposure requires spatially continuous concentration surfaces that also provide reliable uncertainty estimates, yet ground-based monitoring networks remain severely sparse. Kolkata, India’s third-largest metropolitan area (population 14.9 million), exemplifies this challenge: only seven regulatory monitoring stations cover the entire city, leaving large areas unobserved.

This study evaluates how different PM_2.5 surface generation strategies—satellite-based machine learning (ML) and spatial interpolation—differ not only in predictive accuracy but also in their ability to provide decision-relevant uncertainty under sparse monitoring conditions. Using six years of daily observations (2019–2024), we compare two complementary approaches. The first employs satellite-based ML, integrating Sentinel-5P trace gases, MODIS aerosol optical depth, ERA5 meteorological reanalysis, and static urban features (VIIRS nightlights, population density) to predict PM_2.5. The second evaluates spatial interpolation methods—ordinary kriging, inverse distance weighting (IDW), and simple averaging—using station observations alone.

For satellite-based ML (Random Forest), the station-level model achieved R² = 0.79 under leave-one-station-out (LOSO) validation, while grid-based model trained on kriging-interpolated targets reached R² = 0.70 under temporal out-of-sample validation (train: 2019–2022, test: 2023–2024). Feature importance analysis consistently identified dewpoint temperature, air temperature, and surface albedo as dominant predictors, indicating that atmospheric conditions exert stronger control on PM_2.5 variability than emission proxies or land-use variables.

For spatial interpolation evaluated under daily LOSO, all methods achieved comparable point prediction accuracy (R² ≈ 0.85). However, uncertainty calibration diverged sharply. Ordinary kriging achieved 88% empirical coverage for nominal 95% prediction intervals (90% when including observation noise)—approaching theoretical calibration—whereas IDW and simple averaging exhibited severe under-coverage (45–52%), substantially underestimating true prediction error.

These findings yield three key insights: (1) satellite-derived predictors enable spatially complete PM_2.5 estimation beyond monitoring locations, though with moderate accuracy; (2) when temporally aligned station data are available, interpolation achieves higher point accuracy than satellite-based ML; and (3) regardless of estimation strategy, only geostatistical approaches provide uncertainty estimates suitable for health-protective decision-making. We conclude that hybrid frameworks combining satellite-based spatial prediction with kriging-derived uncertainty characterization offer a principled pathway for generating spatially complete and risk-aware PM_2.5 maps in data-sparse urban environments.

How to cite: Raj, A., Dasgupta, T., Sinha, M., and Mitra, A.: Satellite-Based PM2.5 Estimation in Data-Sparse Urban Environments: Comparing Machine Learning and Geostatistical Approaches in Kolkata, India, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-18107, https://doi.org/10.5194/egusphere-egu26-18107, 2026.

Indoor environments account for most human exposure to air pollution, yet indoor air quality (IAQ) monitoring and control remain fragmented across devices, platforms, and proprietary building automation systems. Commercial IAQ monitors and smart thermostats are widely available, but they typically operate in closed ecosystems with limited interoperability. In parallel, open-source communities have demonstrated the potential of low-cost sensing networks, yet these solutions rarely connect to building-level control systems capable of simultaneously reducing pollutant exposure and energy use. To address this gap, we present an open, interoperable framework that integrates open-source and commercial technologies for IAQ monitoring, data management, and automated building control[1]. The framework, developed within the EU-funded healthRiskADAPT project, is built on an open, production-ready IoT infrastructure for indoor environments. At the edge, low-cost sensor nodes collect and transmit environmental data. A web-based interface allows users to register locations, nodes, and sensors, and provides near-real-time visualization, historical analytics, and an interactive map of the sensor network. Beyond monitoring, the framework enables direct integration with commercial control devices such as smart thermostats, smart plugs, and filtration systems. This interoperability supports data-driven control strategies, including increasing ventilation during indoor pollution events, activating filtration during periods of poor outdoor air quality, and dynamically adjusting HVAC operation to balance comfort, energy use, and exposure reduction. By combining continuous mass-balance modeling[2] with real-time sensor data, the system will deliver actionable indoor-outdoor (I/O) ratios and exposure indicators. These outputs could drive automated responses but also support informed user behavior, such as choosing higher-efficiency filters during high-pollution episodes, using kitchen exhaust during cooking, or understanding the trade-offs between energy costs and health risks. In this way, the platform functions not only as a control system but also as an educational and decision-support tool for occupants and building managers. This presentation demonstrates how open-source hardware, open APIs, and modular integration pathways can create a flexible, transparent, and scalable ecosystem for IAQ management. The framework supports diverse use cases, homes, schools, workplaces, and research settings, while offering a roadmap toward energy-efficient, healthier indoor environments driven by interoperable technologies rather than isolated products.

[1] https://particularmatter.org

[2] Dallo, Federico, Thomas Parkinson, Carlos Duarte, Stefano Schiavon, Chai Yoon Um, Mark P. Modera, Paul Raftery, Carlo Barbante, and Brett C. Singer. "Using smart thermostats to reduce indoor exposure to wildfire fine particulate matter (PM2. 5)." Indoor Environments 2, no. 2 (2025): 100088.

How to cite: Dallo, F., Tenti, L., Palo, A., Parkinson, T., and Duarte, C.: Open-Tool Frameworks for Cross-Platform Indoor Monitoring and Optimized Air Cleaning Strategie, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-14158, https://doi.org/10.5194/egusphere-egu26-14158, 2026.

 Background and Objective

Exposures to fine and ultrafine particles (i.e., PM_2.5 and PM₁) are widely accepted as a major environmental risk factor and is known to cause adverse human health outcomes. Most epidemiological research as well as regulatory frameworks rely on using PM_2.5 as the ‘reference’ exposure metric to assess health risks. However, this approach does not adequately quantify size-specific PM effects that are critical for dose-based health assessments. Size-segregated particulate matter assessment of exposures and effects are limited in resource-limited settings. The objective of this study is to estimate lung deposition doses for size-fractionated PM measured using low-cost sensors.

Methods

We utilized the data obtained from an ongoing study conducted in South Indian villages. Here, the household energy use is dominated by biomass combustion and the adoption of cleaner cooking fuels like liquefied petroleum gas (LPG) is relatively low. Ambient PM measurements were carried out continuously over a period of 1 year in 80 rural households in southern India, using real-time, optical, low-cost PM sensors. In order to capture the household-level exposure characteristics, indoor PM measurements were also carried out in a subset of households. Minute averaged, PM mass concentrations in three discrete size fractions: PM₁, PM₂.₅ and PM₁₀ were provided by the low-cost sensors.  The temporal variability in PM concentrations was derived using the time-series data obtained from the sensors. Daily and monthly mean concentrations captured the short-term exposure peaks as well as day to day variability.

Results  

A mathematical model using a non-linear least squares method was developed to transform the measured PM concentrations into a continuous size distribution. Respiratory deposition doses were estimated by feeding the size distribution to a computational model of the lung designed to simulate the spatial and temporal distribution of particles within the human respiratory system, incorporating various deposition models. The estimates of deposition doses ranged from ~0.2µg/min to ~1µg/min in the total lung. The coarse particles contributed to about 20% of the total lung dose, whereas the remaining 80% of the respiratory dose was predominantly of fine and ultrafine particles.

Conclusions

This study demonstrates that physiologically relevant, size-fractioned lung deposition doses can be estimated using limited size-bin data obtained from low-cost sensors. Since low-cost air quality monitoring networks are critical in regions that lack regulatory-grade instrumentation, the proposed analytical framework provides a benchmark for translating low-cost sensor-based air pollution measurements into relevant health-based dose metrics. The proposed analytical framework can be readily modified to incorporate satellite-derived PM inputs alongside low-cost sensor data, enabling improved spatial scaling of size-resolved, dose-relevant exposure estimates.

How to cite: Sai Ganesan, K. D., Puttaswamy, N., Sendhil, S., Natesan, D., Ramasami, R., Desai, M., Pillarisetti, A., Vakacherla, S., Krishnan, R., Sambandam, S., Ramaswamy, P., and Balakrishnan, K.: Estimating Size-Resolved Lung Deposition Doses of Particulate Matter (PM) Using Low-Cost Sensor Data in Rural India, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-6264, https://doi.org/10.5194/egusphere-egu26-6264, 2026.

Traditional solid fuels are extensively used for domestic heating and cooking in developing countries, especially in rural and semi-urban regions. Combustion of these fuels is a major source of nitrogen oxides (NO_x), sulfur dioxide (SO₂), and volatile organic compounds (TVOCs), which significantly contribute to air pollution, respiratory disorders, secondary aerosol formation, and atmospheric photochemical reactions. The generation and release of these pollutants are strongly influenced by fuel moisture content, elemental composition, and inorganic constituents. This study presents a comprehensive investigation of the chemical characteristics of commonly used solid fuels and evaluates the potential of advanced functional materials for mitigating NO_x, SO₂, and TVOC emissions from domestic combustion sources.

Representative fuel samples, including fuel wood (FW), coal balls (CB), dung cake (DC), and crop residues (CR), were obtained from the Raipur–Durg–Bhilai region of Chhattisgarh, India, selected based on their prevalence and area-specific usage patterns. The samples were air-dried, pulverized, and homogenized prior to analysis. Moisture content was determined gravimetrically by oven drying at 105 °C. Ultimate analysis of carbon (C), hydrogen (H), nitrogen (N), sulfur (S), and oxygen (O) was performed using a CHNS/O elemental analyzer. Ionic species, including nitrate, sulfate, chloride, and major alkali and alkaline earth metals, were quantified using ion chromatography to assess their role in pollutant formation and combustion behavior. These chemical parameters were used to infer emission potential for NO_x, SO₂, and TVOCs.

NO emissions were generally higher for AR and DC, while FW showed the lowest NO EF. SO₂ emissions followed a similar trend, with DC producing the highest levels and FW the lowest. TVOC emissions were elevated for fuels with higher moisture and inorganic content, such as AR and DC, whereas FW exhibited the lowest TVOC emission potential. CB displayed intermediate to high emissions, with particularly high TVOC formation due to its variable composition. Emission factors developed in simulated experimental chambers were validated against real-world measurements, indicating that domestic household emissions closely correspond to chamber-based estimates.

To address post-combustion emission control, advanced materials including graphene-based materials, biochar, graphitic carbon nitride (g-C₃N₄), metal oxides (MnO₂/TiO₂), zeolites, metal–organic frameworks (MOFs), covalent organic frameworks (COFs), and silica-based adsorbents were considered for NO_x, SO₂, and TVOC mitigation. Materials were characterized using BET surface area analysis, X-ray diffraction (XRD), Fourier transform infrared spectroscopy (FTIR), and scanning electron microscopy (SEM), confirming high surface activity and strong gas affinity. Their physicochemical properties, high specific surface area, tunable pore size distribution, and surface functional groups, enable efficient adsorption and catalytic transformation of pollutants. Graphene-based materials and biochar adsorb acidic gases through π–π interactions and surface oxygen functional groups, while g-C₃N₄ facilitates photocatalytic oxidation of NO_x under visible light. Metal oxides such as MnO₂/TiO₂ catalyze the oxidation of SO₂ to sulfate and TVOCs to less harmful products via surface redox cycles. Zeolites and MOFs provide selective adsorption of NO_x and TVOCs through microporous confinement and acid–base interactions.

How to cite: Pervez, S., Sahu, D., Pervez, Y. F., Karbhal, I., and Deb, M. K.: Hazardous gaseous pollutants (NOx, SO2, TVOCs) emission from solid fuels combustion and their mitigation using novel adsorbent materials, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-21338, https://doi.org/10.5194/egusphere-egu26-21338, 2026.

The growing urgency of climate action at the city level has led to an exponential rise in documents that describe a city’s policy, action plan, or progress towards climate action. The increased number of documents has made it increasingly difficult for governments to track commitments and compare approaches across jurisdictions. These documents are essential for informed decision-making, but extracting useful information from unstructured PDF reports remains a largely manual, resource-intensive, and inconsistent process. Recent advances in AI and large language model (LLM) based document understanding offer strong potential, but their application in urban climate governance workflows is still limited. Integrating AI-driven document analysis into this workflow offers opportunity for building scalable, standardized, and transparent climate policy assessment.

This study presents an AI-assisted natural language processing (NLP) pipeline that automatically extracts, segments, and classifies climate actions from diverse policy documents. The workflow integrates layout-aware text extraction with an action-segmentation mechanism to identify action statements across heterogeneous formats. A fine-tuned, two-stage ClimateBERT classifier then categorizes actions: Stage 1 differentiates mitigation and adaptation measures (F1 = 93%), while Stage 2 assigns domain-specific sub-categories, achieving 92% F1 for mitigation and 91% for adaptation. An equity-detection module further identifies references to vulnerable groups, inclusivity, and justice-oriented themes.

The pipeline significantly reduces manual review effort and enhances consistency in understanding climate action. By enabling standardized comparisons, the approach directly supports mayors, policymakers, and urban practitioners in evaluating progress and designing more effective and equitable interventions.

As AI capabilities advance, such automated tools will strengthen climate governance by improving the accessibility, reliability, and strategic value of climate policy data.

How to cite: Sahu, S. and Shanker, S. and the MU NLP team: Automated Analysis of City Level Climate Action Plans using Natural Language Processing Technique, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-6386, https://doi.org/10.5194/egusphere-egu26-6386, 2026.

Harmful Algal Blooms (HABs) represent a growing threat to marine ecosystems, aquaculture, and public health in the Central Eastern Arabian Sea (CEAS). This study utilized 18S rRNA metabarcoding to characterize the absolute abundance and community composition of potentially toxigenic diatoms and dinoflagellates in the coastal waters of Goa. The analysis reveals a distinct and alarming prevalence of multiple genera associated with diverse toxin syndromes.

The dataset was dominated by a massive proliferation of the dinoflagellate Karenia (linked to Neurotoxic Shellfish Poisoning and ichthyotoxicity), which reached extreme abundances exceeding 51,000 reads per sample at the most impacted sites. Co-occurring with this bloom, spatially distinct hotspots of Paralytic Shellfish Toxin (PST) producers were identified, specifically Alexandrium and Gymnodinium spp., with Alexandrium counts peaking at over 5,200 reads. Notably, the potent PST producer Alexandrium tamiyavanichii was positively identified, alongside detections of Gymnodinium catenatum.

The diatom community also exhibited significant toxicity potential; the Amnesic Shellfish Poisoning (ASP) genus Pseudo-nitzschia displayed high relative abundance (up to ~3,700 reads), including the presence of P. pungens. Furthermore, vectors for Diarrhetic Shellfish Poisoning (DSP), including Dinophysis spp. and Phalacroma rotundatum, and Yessotoxin producers (Lingulodinium polyedra, Gonyaulax spinifera) were ubiquitously present at lower background levels.

These findings highlight a complex, multi-risk scenario where ASP, PSP, NSP, and DSP vectors coexist within the same coastal system. The distinct spatial separation observed between peak Karenia, Alexandrium, and Pseudo-nitzschia events suggests that heterogeneous environmental drivers are influencing specific HAB assemblages. This data underscores the critical need for broad-spectrum toxin monitoring beyond single-species surveillance in the region.

How to cite: Zedi, S. and Khandeparker, R.: Hidden Hazards in the Central Eastern Arabian Sea: Metabarcoding Reveals Co-occurrence of ASP, PSP, and NSP Vectors in the Coastal Waters of Goa, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-414, https://doi.org/10.5194/egusphere-egu26-414, 2026.

In India, ensuring clean air for all is vital and should not be limited to urbanites. However, the current air quality monitoring networks and clean air strategies are limited to cities. Notably, the air quality status across regions is yet to be measured, and comprehensive regional management plans are non-existent. To address these significant research gaps, the Ambient air quality Monitoring over Rural areas using Indigenous Technology (AMRIT) project was envisioned and implemented by the Centre of Excellence Advanced Technologies for Monitoring Air-quality iNdicators (CoE – ATMAN), Indian Institute of Technology Kanpur. For the first time, over 1,400 low-cost PM_2.5 sensors were installed across the states of Uttar Pradesh and Bihar at the block level. The sensor locations encompass a diverse range of land-use and land-cover categories, demographics, and communities. By leveraging a dense network of low-cost sensors, we developed a data-driven machine learning framework to delineate airsheds for regional air quality management for the first time. We utilized a recurrent neural network–based long short-term memory (LSTM) and hierarchical clustering algorithms with PM_2.5 and meteorological data to delineate airsheds. The LSTM embeddings learn latent representations from PM_2.5–ventilation coefficient (VC) time series, capturing spatiotemporal patterns and inter-variable relationships. These embeddings were then hierarchically clustered to delineate airsheds. Applying this framework, our results for Bihar show five distinct airsheds; three prevail in the north of the Ganges River, and two prevail in the south of the Ganges. Notably, Airshed 1, located in the northwest region, is highly polluted. However, during the post-monsoon and winter across the airsheds, PM_2.5 levels were two to three times higher than the national standard (60 µg/m³), and on ~90% of days, people breathe unhealthy air. Similarly, we identified multiple distinct airsheds in Uttar Pradesh, as well as common airsheds that prevail across Bihar and Uttar Pradesh states. This emphasizes not only shared regional influences but also the need for an integrated approach to reduce PM_2.5 pollution. Therefore, the identified airsheds would be instrumental in targeting the reduction of fine particulate pollution across the Bihar and Uttar Pradesh states. Furthermore, the scalable, data-driven airshed delineation framework using low cost sensors could be implemented across India and potentially globally. Thus, this will facilitate airshed-based air quality management plans and integrated policy interventions to ensure “clean air for all.”

Keuwords: Air Quality; Low-Cost Sensors; Airshed;  Indo-Gangetic Plain; Machine Learning

How to cite: P Chandrasekaran, A., Tripathi, S. N., Godhani, N., Pandey, M., Rai, P., Agrawal, N., Kumar, A., and Ballav, S.: Regional Air Quality Management: A Scalable, Data-Driven Airshed Framework using Low-Cost Sensors across the Indo-Gangetic Plain, India, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-11502, https://doi.org/10.5194/egusphere-egu26-11502, 2026.

Air quality over western India is impacted by both regional emissions and transport from the Indo-Gangetic Plain (IGP). Effective mitigation requires identifying the dominant emission sectors that govern regional air quality during different seasons. In this regard, the present study examines the impact of emissions from major anthropogenic sectors (power, residential, transport and industries) on ambient fine particulate matter (PM_2.5)concentrations over the western Indian region. For this, high-resolution (12km x 12km) regional model (WRF-Chem v3.9.1) simulations have been conducted using the EDGAR v5.0 inventory for anthropogenic emissions. Simulations are conducted for the year 2019, for winter and post-monsoon seasons when PM_2.5 concentrations are typically elevated in this region, and anthropogenic emissions effects are the highest. During winter, the residential sector is seen to dominate, contributing 20% to PM_2.5 concentration over western India, followed by power (9%) and industry (∼8%). The trans-regional pollution from the IGP and central India is also dominated by residential emissions (25%), followed by the power sector (∼8%). In contrast to winter, the dominant source during post-monsoon is the power sector (∼14%), followed by the industry (∼12%) and the residential (∼9%) sectors. Trans-regional impact also shows a similar pattern and dominance of power (∼15%) and industrial (∼10%) sectors. This seasonal shift in the dominant sector is driven by the seasonal variation in emissions. The results also reveal large spatial heterogeneity in sectoral influence, highlighting that dominant emission sources vary at the local scale within western India in both seasons. The study offers model-based insights for more effective planning of regional air pollution mitigation in western India.

How to cite: Shekhar, S., Dhaka, S., Vaishya, A., Ojha, N., Pozzer, A., and Sharma, A.: Sector-segregated anthropogenic impacts on PM2.5 over western India based on regional air quality modeling, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-5792, https://doi.org/10.5194/egusphere-egu26-5792, 2026.