Earth and Environmental Sciences

Abstract Wildfire smoke has become a significant cause of extreme air quality events across the United States. However, its role in the joint occurrence of fine particulate matter (PM2.5) and ozone extremes remains poorly understood. Most prior studies examined the two pollutants separately, even though there is growing evidence that co-occurrence may increase health risks. In this study, we developed a probabilistic spatiotemporal framework to quantify the risk of PM2.5-ozone co-occurring extremes under the influence of wildfire smoke across the contiguous United States from 2004 to 2023. The spatiotemporal Bayesian neural field (ST-BayesNF) model integrates satellite products on smoke coverage, ground measurements of PM2.5 and ozone, and meteorological reanalysis to estimate daily probabilities of co-occurring extremes at a spatial resolution of 5 km. Using a neural field, the model can explicitly capture nonlinear covariate effects, complex spatial dependence, and temporal evolution, as well as providing uncertainty quantification. Model assessments demonstrate that the ST-BayesNF model is able to differentiate coextreme from noncoextreme days and that the model has strong sensitivity to smoke density and its interaction with boundary layer height and near-surface temperature. Predictive results show a significant increase in co-occurrence risk during smoke events, and such amplified risks emerge not only in the western US but increasingly in the Midwest and Northeast due to long-range smoke transport. The ST-BayesNF model provides a probabilistic, spatiotemporally explicit approach for characterizing increased compound extremes under a changing fire regime.

Abstract Overshoot emission scenarios, in which a warming threshold is temporarily exceeded before returning to lower temperatures, are becoming increasingly relevant in the face of insufficient reductions of greenhouse gas emissions. Climate model emulators offer an efficient framework to study these pathways and are seeing growing uptake for this purpose. Yet most existing approaches for spatial emulation are driven solely by global mean surface temperature (GMST). This prevents them from capturing non-reversible local climate responses that persist as GMST declines, raising questions regarding their performance under overshoot trajectories. Here, we consider an aggressive overshoot scenario and evaluate end-of-century biases in mean temperature, maximum daily temperature, and precipitation from a GMST-driven emulator against a 10-member ensemble of simulations from an ESM. The results show that while the emulator displays biases, these biases are largely dominated by internal variability for 99\% of the land regions. Land regions where biases approach or exceed variability are primarily associated with rapid reductions in regional aerosol emissions. We demonstrate that extending the emulator to include aerosol optical depth as an additional predictor helps diagnose and reduce these biases. These results provide evidence that GMST-driven emulators provide useful information for end-of-century projections under overshoot, and that a key priority for future emulator development is to improve the representation of aerosol forcing.

ABSTRACT Extreme precipitation events (EPEs) pose severe risks to society and the economy, and their spatial extent plays a critical role in shaping hydrological impacts. However, it remains unclear whether EPEs of different spatial extents have changed differently and how these changes affect hydrological extremes in the Yangtze River Basin. Here, we identify spatiotemporally contiguous EPEs in the Yangtze River Basin from 1960 to 2024 and classify them as small‐, medium‐, and large‐scale events. While small‐scale EPEs are the most numerous, they have declined significantly in relative frequency. In contrast, medium‐ and large‐scale EPEs have become increasingly common, with the latter becoming more intense. Despite accounting for only 1.1% of all events, large‐scale EPEs dominate annual maximum discharge across 23.3% of river reaches, and their contribution to discharge extremes has increased over time. Precipitation–temperature scaling further reveals that EPE intensity becomes more sensitive to warming as spatial extent increases, indicating a growing risk from widespread events in a warmer climate. These results highlight the spatial extent of EPEs as a key dimension for understanding changing precipitation extremes and basin‐wide flood risk under climate warming.

Pixel-level uncertainty assessment is commonly overlooked in current deep learning-based landslide detection, which greatly limits the reliability, performance, and practical value of the results. To fill this gap, this study employs Monte Carlo Dropout to systematically quantify pixel-level uncertainty. Subsequently, it interprets how external factors, such as terrain, vegetation, clouds, and shadows, interfere with model predictions. Taking SegFormer as an example, this study further investigates how weight allocation across different scales influences the uncertainty. Finally, we introduce an uncertainty ranking-based FP rejection strategy coupled with an FN priority capture strategy to improve the efficiency of mapping results inspection, thus improving landslide identification performance. The results indicate that by removing only the top 10% of pixels with the highest uncertainty, the mIoU increases by at least 7%. This study greatly enhanced the reliability, performance, and practical value of remote sensing-based landslide mapping from a new perspective.

Abstract. The White Nile from Lake Victoria through Lakes Kyoga and Albert to the Sudd forms a complex lake-river-wetland corridor where flood propagation, storage, and attenuation remain poorly quantified. Following unprecedented and persistent flooding across South Sudan in 2022, this study quantified how long it takes a flood wave to travel from Lake Victoria to the Sudd and how upstream storage and connectivity shape multi-year flood behaviour. Using daily lake levels, discharge, CHIRPS rainfall, and MODIS-derived inundation for 2002–2024, we tracked sequential flood peaks through the Victoria–Kyoga–Albert–Sudd cascade and mapped monthly wetland dynamics across five South Sudan sub-catchments. Flood-wave tracking showed a mean system transit time of 16.84 ± 1.95 months (range 13.0–20.9 months), overturning the long-held assumption of a five-month propagation. Segmental analysis revealed rapid transmission from Victoria to Kyoga (mean 4.2 months) but strong attenuation through the Albert–Sudd reach (mean 9.3 months), consistent with extensive floodplain storage and backwater control. Correlations between Lake Victoria peaks and downstream wetland extents strengthened markedly after 2019, with r² exceeding 0.8 at 9–13-month lags, confirming strong hydraulic coupling and long system memory. The 2019–2024 high-water regime was therefore not a series of isolated rainfall events but a multi-year propagation of excess storage initiated by the 2019 positive Indian Ocean Dipole anomaly and consecutive rainfall seasons. When compared with historical episodes in the 1870s and 1960s, the persistence and spatial reach of the 2019–2024 floods rank among the most extensive in the modern record. These results redefine the White Nile as a long-memory system where upstream storage governs downstream flood risk, offering a new empirical basis for flood forecasting, wetland management, and anticipatory action in South Sudan and the wider basin.

Extreme drought events have become increasingly frequent under ongoing climatic change, thereby constraining vegetation growth and altering ecosystem processes. Vegetation recovery time following drought plays a crucial role in ecosystem stability, and extensive studies have been conducted to quantify vegetation recovery. However, most studies estimate vegetation recovery time using a single vegetation index, which does not adequately reflect how vegetation responds to drought conditions, since different vegetation indicators reflect different facets of vegetation dynamics. In this study, multiple vegetation indicators, including Normalized Difference Vegetation Index (NDVI), Enhanced Vegetation Index (EVI), Leaf Area Index (LAI), Gross Primary Productivity (GPP), and Solar-Induced Chlorophyll Fluorescence (SIF), were applied to investigate post-drought vegetation recovery in the Yangtze River Basin (YRB). The findings reveal that: (1) most vegetation recovered within four months after drought, with one-month recovery being the most prevalent, followed by four-month recovery; (2) the average recovery times derived from EVI, LAI, NDVI, GPP, and SIF were 2.23, 1.62, 2.45, 2.79, and 2.00 months respectively; (3) forests exhibited the fastest recovery rates, whereas shrublands recovered the slowest. This study assesses post-drought vegetation status via the recovery duration, offers theoretical basis for water resource allocation optimization, and provides important reference for coping with future ecological risks.

Surface deformation induced by underground coal mining is characterized by strong nonlinearity and spatial heterogeneity, which complicates early warning and ecological assessment. While InSAR-driven data assimilation models and remote sensing-based ecological indices are widely used for long-term monitoring, three fundamental limitations remain unresolved: (1) severe spatial imbalance in deformation samples biases data-driven models toward mean-reverting predictions, (2) recursive multi-step forecasting accumulates errors, leading to instability in long-horizon extrapolation, and (3) in ecological monitoring, vegetation resilience further induces a multi-year observation lag, resulting in a “pseudo-stable” bias in optical indicators. To address these issues, this study proposes an unified framework integrating multi-step deformation prediction and ecological time-lag analysis. Taking the Datong Coalfield as the study area, we utilized 231 Sentinel-1A images from March 2017 to December 2024 for SBAS-InSAR deformation inversion. A spatial stratified sampling strategy is used to extract 5894 representative points. A 24-step backward and 15-step forward windows were reconstructed to systematically compare six predictive models. Simultaneously, the Remote Sensing Ecological Index (RSEI) derived from Landsat data is used for cross-lagged analysis. The results demonstrate that: (1) The maximum deformation rate reached −276.75 mm/year, with cumulative subsidence exceeding −2000 mm. (2) At 3-step short-term forecasting, all models proved robust, with LSTM performing best (RMSE = 5.78 mm). At 15-step extreme extrapolation, however, traditional recursive models diverged significantly (Kalman, RMSE = 45.70 mm), whereas N-BEATS maintained stability and effectively mitigated temporal error cascades with an RMSE of 17.98 mm. (3) The core collapse zone exhibited concurrent ecological degradation (Lag 0), while the marginal basin presented a hidden degradation period of one to two years. It provides reliable scientific support for precise tracking and proactive safety management in complex mining areas.

Abstract. Runoff threshold behavior is widely reported in event-based hydrological studies, but its interpretation and cross-catchment variability remain unresolved because threshold metrics, values, and process interpretations vary among studies, climates and landscape settings. This study synthesizes reported storm-runoff thresholds from 138 experimental catchments worldwide, as well as reported dominant runoff mechanisms, documented wetness-dependent mechanism transitions, and soil-geology-hydrogeology associations. Across the reviewed literature, threshold-like responses were identified using rainfall metrics (e.g., event rainfall amount and rainfall intensity), hydrological-state metrics (e.g., antecedent or within-event soil moisture, storage, and groundwater level), and composite rainfall–state indicators. Hydrological-state and composite indicators were reported more frequently than rainfall-only metrics. Subsurface- and saturation-related mechanisms were most frequently reported, particularly among studies in humid catchments. Among the catchments with explicitly documented event-scale transitions in runoff generation mechanisms, shifts from surface-dominated responses toward saturation-, subsurface-, or shallow-groundwater-influenced responses were more frequently reported as catchment wetness increased within the reported transition subset, although reverse and context-dependent pathways are hydrologically possible. Co-occurrence analysis indicates that reported mechanisms are associated with soil-depth, texture, permeability, lithology, and hydrogeological descriptors, which we interpret as structural contexts that condition state-dependent functional connectivity. Together, the synthesis supports a connectivity-based framework in which rainfall forcing interacts with catchment state and structural constraints to activate or connect runoff pathways.

Since the implementation of the Grain-for-Green Program (GGP), vegetation across the Loess Plateau (LP) has substantially recovered. However, whether the associated increase in ecosystem carbon gain was accompanied by a proportional increase in water consumption and whether groundwater storage changed synchronously remain unclear. This study integrated multi-source remote sensing products, GLDAS-Noah land-surface assimilation data, GRACE/GRACE-FO satellite gravimetry, irrigation water-use data, provincial water-use statistics, and coal-resource information to examine long-term changes in gross primary productivity (GPP), evapotranspiration (ET), water-use efficiency (WUE), soil moisture (SM), and groundwater storage anomaly (GWSA) during 2002–2023. GPP increased significantly by 10.67 g C m−2 yr−1 (p<0.01), whereas ET increased more modestly by 1.97 mm yr−1 (p<0.05). The relative growth rate of GPP (1.66%) was approximately 3.5 times that of ET (0.47%), and WUE increased by 0.018 g C m−2 mm−1 yr−1 (p<0.01). In the XGBoost–SHAP models for 2004–2019, LAI showed the strongest model-based association with GPP and WUE, whereas ET was associated more broadly with LAI, air temperature, and precipitation. SM declined during 2002–2015 but showed an increasing tendency during 2016–2023, particularly in the middle and deep layers. The long-term GWSA slopes derived from CSR and JPL were −8.707 and −9.505 mm yr−1, respectively, and the averaged CSR–JPL GWSA series showed a Sen’s slope of −9.131 mm yr−1. GWSA declined during 2002–2020 and showed only a short-term, nonsignificant increase during 2020–2023 (4.110 mm yr−1, p>0.05). These contrasting trajectories indicate that increases in surface carbon uptake and improvements in soil-water conditions were not accompanied by synchronous regional groundwater recovery. Overall, the ecological-restoration period was accompanied by increased carbon gain and WUE without a proportional increase in regional ET, while groundwater storage followed a distinct trajectory. These findings provide regional-scale evidence and a quantitative basis for coordinating sustainable water-resource management with ecological-restoration optimization on the LP.

Abstract The seasonal migration of the planetary-scale convectively coupled system of Intertropical Convergence Zone brings sustained rainfall over the Indian subcontinent and establishes large-scale monsoon. However, to make accurate predictions that support agricultural policies, we must define the onset of sustained rainfall at every location. The necessity of defining the onset of a large-scale phenomenon at a local-scale appears elusive and remains unresolved. Operational definitions are unable to differentiate between large-scale and local onsets, thereby restricting their utility to farmers. We propose a novel trans-disciplinary perspective that separates onsets due to local transient phenomena from true onsets related to the large-scale monsoon system. We view the arrival of monsoon as the emergence of a large-scale persistent organization from the spatiotemporal evolution of clusters of local onsets. Using network science, we track the largest cluster formed by connecting locations in spatial proximity that have undergone local onset. We then define the large-scale onset at one location when that location becomes part of a planetary-scale cluster of onsets. Thus, we establish monsoon as a phase transition in the framework of climate networks. We also discover that climatological monsoon progression over India comprises two abrupt growths in the size of the largest cluster of local onsets delineating the emergence and spatial spread of Indian monsoon. Climatologically, the large-scale onset over peninsular India occurs after a large-scale cluster is established over Northeast India (NEI). The cluster from NEI propagates west and merges with a cluster over the peninsula, establishing a planetary-scale monsoon system that progresses northwards consistently.

Abstract El Niño events exert a profound influence on the global carbon cycle by imposing widespread heat and water stress on terrestrial ecosystems. Plant isoprene emissions respond rapidly to such stress, yet it remains unclear whether this response can track the spatiotemporal evolution of El Niño’s impacts over land. Here, we used satellite-derived global isoprene emissions for the first time to assess the west-east progression of the 2015–2016 El Niño. We observed that isoprene emissions increased by up to approximately 30% across tropical ecosystems relative to the climatological mean, with pronounced anomalies emerging during the event. The spatiotemporal evolution of these anomalies closely aligns with the El Niño progression inferred from sea surface temperature (SST) anomalies in the equatorial Pacific. In contrast, commonly used satellite vegetation products, including leaf area index (LAI) and solar-induced chlorophyll fluorescence (SIF), show weaker and spatially incoherent responses. These results demonstrate that satellite-derived isoprene provides a sensitive and mechanistically grounded tracer of ecosystem stress, offering a complementary perspective for monitoring the intensity and propagation of extreme climate events across terrestrial ecosystems.

Abstract Environmental baselines for deep-seabed mining must be sufficient to separate the effects of mining from a climate background that is itself changing. We used a five-model Coupled Model Intercomparison Project Phase 6 (CMIP6) ensemble to ask when climate-driven change in the abyssal Clarion–Clipperton Zone becomes distinguishable from natural variability — its time of emergence — for temperature, dissolved oxygen and pH at 5000 m, and for particulate organic carbon export at 100 m as a proxy for food supply reaching the seafloor. We expressed the displacement of a ten-year running mean from a modelled 2015–2030 reference state as a signal-to-noise ratio (SNR) against preindustrial control variability; emergence required an SNR above 2 sustained for five consecutive years. Robust emergence was rare by 2035, occurring in one model and for temperature only. By 2050 it had begun in 17 of 34 model–variable–scenario combinations and been confirmed in 16. The changes at emergence were physically small: oxygen shifted by less than 1% of its background concentration and temperature by a few thousandths of a degree, well below plausible physiological thresholds. Food supply was the exception, with a 15% decline in the single case that emerged — one model, under the high-end SSP5-8.5 pathway only, and not before the late 2050s. Ten-year records proved too short to characterise deep-ocean variability, and Areas of Particular Environmental Interest differed from contract areas once depth and location were matched. The variables that emerge earliest are therefore those least likely to matter biologically, while the variable most likely to matter is one whose upper-ocean supply mining does not set, though mining can alter that supply in transit and at the seabed. That is where attribution will be hardest, and why a fixed historical baseline needs to be supplemented rather than relied upon.

Abstract. Recent studies have highlighted that flooding can influence population dynamics. However, existing estimates of future fluvial flood damage primarily consider population changes driven by births, deaths, and migration unrelated to flooding. As a result, the potential impacts of flood-induced population movements (FIPMs) on future fluvial flood damage costs remain largely unexplored. This study evaluated the impacts of FIPMs on future fluvial flood damage costs in Japan, a country that faces flood risk and population decline. We developed a methodological framework that uses statistical causal inference to quantify FIPMs, integrates these estimates into future population and land-use projections, and evaluates future fluvial flood damage costs under scenarios of climate and land-use change. Empirical relationships between flood magnitude and FIPMs were estimated using grid-cell-level data from two flood events in 2019 and 2020 and municipality-level data from municipalities affected by flood disasters during 2015–2020, and then applied to future population projections using flood magnitude indicators derived from future fluvial flood inundation analyses. The results indicate that incorporating FIPMs leads to only modest changes in estimated fluvial flood damage costs at the national level (generally below 1 %), and similarly modest impacts at the prefectural level, except for a few prefectures with changes of approximately 2 %. However, greater variability is observed at the municipal level, with approximately 10 % of municipalities experiencing changes exceeding 1 % and some municipalities showing reductions in estimated fluvial flood damage costs exceeding 10 %. These findings highlight the importance of accounting for FIPMs in municipal-level fluvial flood risk management frameworks and policy evaluations.

Climate change and ongoing urbanization are increasing human exposure to heat stress in cities, where strong spatial contrasts in shading and imperviousness generate large variations at the scale of streets and public spaces. Urban adaptation planning therefore requires high-resolution information on outdoor heat stress. Microscale climate models provide such information, but their reliability must be demonstrated at meter-scale spatial resolution. This study presents a validation approach for such models through a detailed evaluation of the coupled UrbClim-HiREx modeling framework. UrbClim is a fast urban boundary layer model operating at 100 m resolution, while HiREx is a radiation-based model that simulates wet-bulb globe temperature (WBGT) at 1 m resolution. Model performance is assessed against a dense observational dataset collected during 7 consecutive heat wave days in June 2023 in Ghent, Belgium, spanning multiple urban microenvironments within a 300 m × 300 m area. UrbClim reproduces air temperature and humidity with high accuracy and provides realistic pedestrian-level wind speeds after in-canopy extrapolation, yielding a reliable forcing for HiREx. HiREx reproduces the diurnal evolution of WBGT with root mean squared errors of 0.95 °C at an open site and 0.51 °C beneath tree canopy. It resolves transient shading effects and meter-scale spatial variability, with hourly daytime root mean squared errors across the microenvironments remaining below 1.59 °C. The model reproduces the observed reductions in peak daytime heat stress under tree shade of approximately 4 °C WBGT. These findings provide the most detailed validation of the UrbClim-HiREx modeling framework to date and establish a transferable approach for evaluating microscale heat stress models. The results confirm the framework’s suitability for identifying urban heat stress hotspots and assessing adaptation strategies at meter-scale resolution.

The rime-splintering (RS) process, which is one of the most important secondary ice production mechanisms, plays a crucial role in regulating ice crystal number concentrations and precipitation formation in convective clouds. The initiation and efficiency of the RS process are modulated by the spatial distributions of liquid and ice particles, which are highly heterogeneous. However, numerical models are developed based on the idealized assumption of homogeneous liquid-ice mixing in a given volume, which causes large uncertainties in modeling ice production in convective clouds. Based on in-situ measurements from The Convective Precipitation Experiment (COPE), this study investigates how the heterogeneous particle distribution affects the RS process by defining an impact factor F HM . The liquid-ice mixing homogeneity (χ) in a given cloud is quantified using information-theoretic entropy, and the spread of riming rate (Δ R rim ) is quantified using the maximum R rim minus the minimum one. The results show that F HM increases with χ at a given temperature, indicating that a greater liquid-ice mixing homogeneity enhances the occurrence of RS process. Since the RS process is non-linear, in most of the observed clouds F HM exceeds 1, and increases with increasing Δ R rim , suggesting the heterogeneous particle distribution enhances the mean ice production rate compared to that assuming homogeneous particle distribution. Among the various meteorological factors that influence χ and Δ R rim , vertical wind shear (WS) and condensed water content (CWC) show the strongest yet opposing correlations with χ, while temperature (T) and WS contribute the most to Δ R rim . These insights provide a new perspective for advancing subgrid-scale parameterization of the RS process.

Over the past two decades, changes in sea temperature and sea-ice conditions have driven profound ecological shifts along the coast of East Greenland. One of the most conspicuous developments has been the recent and widespread occurrence of boreal baleen whale species in coastal waters. Fin whales, humpback whales, and minke whales now regularly feed along the East Greenland coast, at least as far north as 70°N and likely beyond, whereas historically, these species were only rarely observed in the region. Systematic surveys conducted in 2015 and 2024 indicate regional abundances of approximately 4,000 humpback whales, 6,000 fin whales, and 4,500 minke whales in the region. Capelin have been identified in stomach contents of minke whales from East Greenland, and both fin whales and humpback whales are likely to feed on capelin and krill in the region. The combined prey consumption of these baleen whales during their approximately four-month seasonal residency in East Greenland is estimated at about approximately 1 million tonnes annually, a substantial proportion of which may consist of capelin and krill. At the same time, recent observations suggest a summer redistribution of capelin from Icelandic waters toward East Greenland, consistent with rising sea temperatures around Iceland and along the East Greenland shelf. Oceanographic records show a clear change point in East Greenland sea surface temperatures in the late 1990s, followed by an average increase during the summer months of approximately 1 °C. The summer sea-ice regime shifted markedly after 2000, reducing the extent of pack ice along the coast. Together, these physical changes have facilitated access of boreal baleen whales to coastal East Greenland waters and may have enabled exploitation of newly established capelin resources. These developments indicate a rapid borealization of the East Greenland marine ecosystem, with potentially major implications for pelagic food webs and predator–prey dynamics in the region.

Abstract Coupled climate models generally simulate a decline in Antarctic sea ice extent (SIE) over the satellite period, failing to capture the observed near‐zero trend. Recent work has suggested that high‐resolution ocean models could ameliorate this issue via improved representation of mesoscale ocean processes, which may delay simulated Antarctic SIE decline. We investigate this hypothesis using two high‐resolution hierarchies of coupled climate models developed at GFDL and NCAR, which span atmospheric resolutions of 0.25–1 and ice‐ocean resolutions of 0.1–1, as well as three generations of Coupled Model Intercomparison Project (CMIP) models. Across the high‐resolution hierarchies and CMIP ensembles, we find no clear relationship between ocean and atmospheric resolution and Antarctic SIE trends over both the satellite period and the remainder of the 21st century. Climate models that reproduce observed trends in individual ensemble members are those that exhibit large amplitude multi‐decadal Southern Ocean climate variability.

The Greater Cape Floristic Region (GCFR) in South Africa is a fire-prone biodiversity hotspot where high species richness, structural complexity, and small plant sizes (0.0001–4 m2) pose substantial challenges for remote sensing-based biodiversity assessment. Spectral similarity among species and the mismatch between plant size and sensor pixel dimensions limit the capacity of current and forthcoming spaceborne systems to resolve individual species and accurately detect plot-level diversity changes. We therefore developed a physics-based simulation framework that couples fynbos trait measurements with radiative transfer modeling in the DIRSIG (Digital Imaging and Remote Sensing Image Generation) environment towards quantifying information loss across spectral and spatial scales and to define theoretical limits for biodiversity monitoring. We constructed a three-dimensional virtual scene of post-fire fynbos communities in Grootbos Private Nature Reserve, integrating high-resolution imagery, terrestrial laser scanning (TLS), and structure-from-motion (SfM)-derived point clouds. Field measurements of mean diameter and percent cover were used to scale vegetation models and constrain species abundance. We distributed plant instances using a blue noise sampling algorithm, guided by density maps derived from unmanned aerial system (UAS) imagery. Species-specific optical properties were parameterized using field-measured reflectance data and the PROSPECT radiative transfer model, while terrain structure was derived from SfM-based digital terrain models. The integrated scene was used to simulate multispectral (DJI Mavic 3 MSI), hyperspectral (AVIRIS-NG), and light detection and ranging (LiDAR) observations. Agreement between simulated outputs were evaluated against corresponding field-acquired datasets using spectral signatures and vegetation indices. This framework enables systematic assessment of sensor specification effects on spectral biodiversity metrics and provides a pathway for evaluating theoretical limits of species discrimination across airborne and satellite platforms.

Peatlands, critical global carbon sinks, are facing increasing threats from climate change-driven heatwaves and droughts. These threats can cause a midday depression in carbon uptake through photosynthetic inhibition. Using high-temporal-resolution solar-induced chlorophyll fluorescence (SIF; ~30 s) and chamber-based CO2 flux measurements, we investigated the coupling between SIF and gross primary production (GPP) during extreme events (air temperature > 25 °C and vapour pressure deficit > 15 hPa) in a northern peatland. Our results show that SIF tracks GPP closely under non-stress conditions (daily R2 = 0.86–0.96). However, during combined heat and drought stress, midday correlations collapsed (Case A: R2 = 0.04 on 27 June; Case B: R2 = 0.15 and 0.01 on 29 and 30 June, respectively), indicating severe decoupling. Importantly, we discovered legacy effects from multi-day heat exposure: on 26 June, vegetation with prior cumulative stress (Case A) showed weak morning coupling (R2 = 0.07), while vegetation without prior stress history (Case B) maintained strong coupling (R2 = 0.93). This suggests that cumulative stress alters baseline physiology and can exacerbate midday mismatches; therefore, not just current condition controls photosynthetic regulation. These findings highlight limitations of SIF-based GPP estimation at sub-daily timescales during stress, particularly in heterogeneous peatland systems where canopy composition and physiological responses could vary among plant functional types.

Abstract. Integrated Assessment Models (IAMs) serve as critical instruments for scenario-based analysis and have been instrumental in informing environmental policy at both global and regional scales. However, their limited geographical and sectoral scope constrains their ability to evaluate comprehensive policy packages such as the European Union's Fit-for-55. GCAM-Europe, an expansion of the well-stablished Global Change Analysis Model (GCAM) addresses this gap by explicitly representing energy, land use, agriculture, water, and emissions systems for European Member States and key non-EU countries. Operating within a global framework, the model enables integrated assessment of policy impacts both across and within European regions, while also capturing spillover effects in other regions in the world. As an open-access and continuously evolving platform, it provides a valuable tool for European researchers, policymakers, and stakeholders to design, test, and evaluate climate and environmental strategies that support a just and effective climate transition.

Abstract. Microplastics (MPs) are environmental contaminants of global concern. Although the relevance of the atmosphere in the transport and distribution of MPs worldwide has been acknowledged, country-scale quantitative data on wet and dry MP deposition rates remain limited. We therefore quantified MPs in wet and dry atmospheric deposition samples collected on a four-weekly basis over a one-year period between May 2024 and May 2025 at one urban (Zurich), one suburban (Duebendorf), two rural (Magadino and Payerne) and one mountainous site (Chaumont) in Switzerland. We used focal plane array μ-Fourier transform infrared spectroscopy to identify MPs in the 20–215 µm size range and included a rigorous assessment of the measurement uncertainties. Particle sizes were converted into masses to obtain mass deposition rates. The number- and mass-based MP deposition rates were highest at the urban site, with respective means of 881 MPs m−2 d−1 [95 % confidence interval (CI): 562–1199] and 53 µg m−2 d−1 [CI: 17–107]. The deposition rates were lower and similar among the remaining sites, ranging from 249 to 331 MPs m−2 d−1 [CI: 140–478] and from 13 to 21 µg m−2 d−1 [CI: 4–46]. Based on the determined deposition rates and land-use statistics, an annual deposition of 219 t or 3.8×1014 particles was estimated for MPs of the analyzed 20–215 µm size fraction excluding tire wear particles, in regions <2000 m above sea level across Switzerland. Corresponding annual atmospheric inputs of MPs to Swiss agricultural land and surface waters were estimated at 78 and 10 t, respectively.

Abstract. As weather-related disasters become more frequent and severe, there is a growing global push toward impact-based early warning systems, exemplified by initiatives such as EW4All. This transition positions machine learning (ML) and artificial intelligence (AI) as powerful tools for integrating meteorological hazard data with information on vulnerability and exposure into data-driven forecasting systems. In this work, we explore the use of 112 emergency calls as high-resolution impact proxies for an ML-based prediction problem. Specifically, we develop a model that combines rainfall-related weather data, information on recent emergencies, and static vulnerability-exposure layers to predict, at a municipal and hourly resolution, whether flood-related impacts will occur in the next hour. This study covers the period between October 2018 and February 2025 in Catalonia, northeastern Spain. To address the severe temporal class imbalance and uncertainty characteristics of emergency call data, we define a custom walk-forward evaluation scheme that ensures the same number of positive samples across comparable time periods. We then distribute municipalities into three distinct population density groups (low, medium, and high) and train one model for each one. This stratification enables us to evaluate performance across diverse population dynamics and varying data availability. The resulting models are compared against operational methodologies, such as climatology-based weather warnings issued by meteorological agencies. Our results show that the ML approach represents a substantial improvement in two out of the three groups. The model for the lowest-density group, however, struggles due to a substantial lack of impact data, highlighting a key roadblock for data-driven algorithm development in sparsely populated regions. Further experiments focus on explainability to improve trust by understanding the models' behaviours using feature importance analyses like SHAP (SHapley Additive exPlanations), feature group selection, and ablation. Additionally, we present a methodology to evaluate model behaviour across different stages of a rainfall event. These experiments show how ML models can effectively combine various data sources to provide more effective predictions than using these sources in isolation. Moreover, while the best-performing model exhibits moderate performance at the onset of rain, it still consistently improves upon traditional baselines. As the event evolves, its prediction skill rises significantly and is maintained even after precipitation has stopped. This highlights the strong potential of even relatively simple ML pipelines to deliver timely, localised anticipation of weather-related impacts.

The Tibetan Plateau (TP) is characterized by frequent mixed-phase clouds, in which riming exerts a crucial control on precipitation formation. However, conventional bulk microphysics schemes generally ignore the regulatory impact of sub-grid liquid-ice mixing state on riming processes. This study improves the riming parameterization within the WRF Morrison scheme by adopting constant reduction coefficients and introducing a diagnostic mixing homogeneity parameter χ for snow and graupel, which varies with condensed water content and temperature. Reduced riming coefficients correct simulated precipitation distribution and spatial deviation, while χ -based dynamic regulation further optimizes temporal variation of regional precipitation, reducing RMSE by 23.8%. The modified scheme shows superior performance across all rainfall grades, particularly for heavy and torrential rain. Physically, χ inhibits excessive riming, maintains supercooled liquid droplets, and effectively alleviates the overestimation of surface precipitation. One-month simulations validate that considering liquid-ice mixing homogeneity markedly improves WRF precipitation simulation capability over the TP.

Chlorophyll-a concentration is a key indicator reflecting the growth status of phytoplankton, and its accurate prediction is of great significance for assessing the degree of water eutrophication. Although existing approaches have achieved good performance, they generally pay insufficient attention to multi-scale spatial information and show limitations in characterizing the continuous spatiotemporal dynamics. To address these issues, this paper proposes a multi-scale spatiotemporal graph ODE network (MGODE) for ocean chlorophyll-a prediction. The MGODE adopts a dual-layer structure, simultaneously processing chlorophyll-a concentration data at both the region level and node level to capture multi-scale spatial features, and it enables effective interaction of cross-scale features through dynamic transmission coefficients and a gated fusion mechanism. Meanwhile, the MGODE employs a dual-ODE architecture at both the node and region levels, utilizing spatiotemporal ODE blocks to continuously and deeply capture features, thereby simulating the continuous spatiotemporal dynamic evolution of chlorophyll-a. Experiments on real-world datasets from the Bohai Sea and South China Sea show that the proposed MGODE model achieves higher prediction accuracy than several current state-of-the-art models. Compared with the best baseline, the MGODE achieves reductions of 2.78% in MAE and 1.07% in RMSE on the Bohai Sea dataset and reductions of 1.19% in MAE and 1.38% in RMSE on the South China Sea dataset. These results demonstrate the potential of the MGODE to support marine chlorophyll-a forecasting and marine ecological monitoring.

ABSTRACT Infrastructure resilience to environmental change is critical for maintaining the delivery of essential services to our society, including power generation, water supply, transport and telecommunication. Computational modelling has become integral to infrastructure planning and management, enabling the exploration of system interdependencies and the testing of adaptation strategies against unprecedented conditions. However, infrastructure model outputs are conditional on a range of uncertain assumptions about the system drivers and properties, both in the present day and in the future under climate change. In this article, we show how global sensitivity analysis can be used to consistently quantify and attribute the uncertainty in infrastructure model outputs as a consequence of input uncertainties, using two examples from the energy and water sector in the United Kingdom (UK). We find that dominant uncertainties vary case by case, as does the relative importance of climate uncertainty. We argue that structured uncertainty and sensitivity analysis should be incorporated more consistently across infrastructure modelling sectors to support resilient infrastructure design.