Earth and Environmental Sciences

Study region : Asia Pacific region, Taiwan. Study focus : Accurate discharge estimation in river systems is fundamentally hindered by the scarcity of high-flow measurements and the complexities of channel non-stationarity. This study introduces an integrated framework that reconciles historical in-situ observations with hydrodynamically derived synthetic relationships to bridge critical data gaps. We employ a dual-methodology paradigm, utilizing both a Bayesian inference model (BaRatin) and a fuzzy regression framework to simultaneously address probabilistic variability and inherent measurement uncertainty. This approach ensures a comprehensive evaluation across the full spectrum of flow regimes at two gauging stations in the Wu River Basin. New hydrological insights : This research explicitly incorporates high-flow extensions generated via the Height Above the Nearest Drainage (HAND) method into the uncertainty assessment, offering a physically consistent approach that utilizes hydraulic constraints rarely explored in rating-curve literature. Results demonstrate that while uncertainty is most pronounced under low-flow conditions due to channel irregularities, the bounds converge significantly to within ± 35%–75% in high-flow regimes. The Bayesian method effectively captures structural errors in complex channel–floodplain reaches, whereas fuzzy regression offers a computationally efficient alternative for data-scarce scenarios. Notably, integrating HAND-based hydrodynamic information reduces high-flow bias and enhances the continuity between empirical and synthetic segments. This framework establishes a statistically defensible basis for discharge estimation, significantly improving uncertainty quantification in both gauged and ungauged catchments.

Enhanced rock weathering (ERW) is an emerging carbon dioxide removal (CDR) pathway that has been heavily explored for agricultural systems. There may also be co-benefits from this approach in historically acidified forest ecosystems by replenishing depleted nutrient cations, leading to increased tree growth. However, far less analysis has been conducted for forest land applications. Process-based modeling of these benefits is currently preferable because there are insufficient empirical data on ERW in forests to effectively capture the spatial heterogeneity in climate, soils, and geology across the U.S. We developed and calibrated a mechanistic model that simulated a single 40 tonne per hectare basalt application rate across 132 million hectares of practically manageable forested land within the conterminous United States. The gross CDR potential is 660 Mt CO 2 e from a single application, annualized to 66.0 Mt CO 2 e yr −1 over 10 years. Abiotic ERW was geographically dispersed and accounted for 76.6% of the total, while biotic ERW accounted for 23.4% and was clustered in the Pacific Northwest and the East. After carbon penalties were assessed, using a life cycle assessment, the net CDR potential is 353 Mt CO 2 e from a single application (35.3 Mt CO 2 e yr −1 over 10 years), with opportunities to increase net sequestration through reapplication. In this study we provide the first detailed assessment of the potential for ERW in U.S. forests under realistic supply-chain assumptions.

Spatially explicit information on forest resources and structure is essential for sustainable forest management and evidence-based policy-making. In the Nordic region, large-scale forest mapping often relies on integrating National Forest Inventory (NFI) field plots with airborne laser scanning (ALS) data. However, infrequent nationwide ALS campaign coverage limits their use for continuous monitoring. Satellite imagery, with its high temporal and spatial resolution, provides a promising alternative. We evaluate UNet-based deep learning models trained on wall-to-wall ALS-derived forest resource maps for predicting volume and Lorey’s height in Norway using optical (Sentinel-2) and SAR (Sentinel-1, PALSAR-2) data. The UNet models, trained on both Finnish and Norwegian ALS maps, are benchmarked against extreme gradient boosting (XGB) models. Transfer learning is further explored by finetuning models using Norwegian NFI plots. Model accuracies are assessed using 541 reserved test NFI plots and 44 independent forest stands, representing high‑volume mature boreal forests (>200 m 3 ha −1 ). The UNet model trained on Norwegian ALS‑based data achieved R2 values of 0.59 for both volume and Lorey’s height when evaluated on NFI plots, and 0.70 and 0.59 for forest stands, respectively, outperforming the XGB models. Finetuning improved model transferability, yielding gains of up to 0.13 in R2 for volume and 0.46 for Lorey’s height when adapting the Finnish model to Norwegian conditions. Utilizing SAR data alongside optical data enhanced model accuracy. Overall, our findings demonstrate the potential of UNet models trained on wall-to-wall ALS maps for forest resource mapping across Nordic countries.

Hyperspectral anomaly detection aims to find abnormal targets without prior information. However, the high dimensionality of hyperspectral data, complicated spatial structures, and varying object scales make it challenging to jointly utilize spectral and spatial information. Therefore, anomalies may be confused with background regions. A Balanced Spectral–Spatial Cross-Fusion Network (BSCF-Net) is proposed for hyperspectral anomaly detection. The network uses a multi-branch encoder, where spatial branches capture features with different receptive fields and spectral branches extract spectral patterns through one-dimensional convolutions and channel attention. The Bidirectional Spectral–Spatial Cross-Attention (BSCA) mechanism enables information exchange between spectral and spatial features. The Multi-Scale Gated Refiner (MSGR) module is used to refine the fused features. With an autoencoder reconstruction framework, BSCF-Net identifies anomalies according to reconstruction errors and reduces background interference. Experimental results on five public hyperspectral datasets demonstrate the effectiveness of BSCF-Net, achieving competitive AUC performance under diverse background conditions.

Abstract Secondary organic aerosol (SOA) formed from wildfire/biomass-burning emissions (BB) represents a significant fraction of global SOA production. However, there are large uncertainties in representing BB-SOA in climate models. We studied the evolution of organic aerosols (OA) from burning three biomass samples─savannah grass, savannah wood, and boreal forest surface─under different combustion conditions and during daytime (photochemical oxidation) and nighttime (dark oxidation) aging in an atmospheric chamber. OA dominated the BB emissions by contributing ∼82–99% of the total PM1 mass. Atmospheric aging by both oxidation processes produced comparable amounts of net OA mass. We show, with PMF analysis, that this is connected to the more efficient loss of primary OA, which compensates for the more efficient gas-phase oxidation during daytime aging compared with nighttime. The observed moderate OA mass enhancement (0.75–1.3 times) agrees well with field observations, thereby addressing the discrepancy between laboratory and field studies. For both aging scenarios, total OA emission factors after aging are similar to those of primary OA, providing new insights into the evolution of BB emissions. Our results suggest a simplified treatment of OA in climate models in remote areas with low NOx concentrations.

Mangroves are important coastal blue-carbon ecosystems, and accurate biomass estimation is essential for carbon stock assessment and ecological monitoring. To address the limitation of regional-scale biomass estimation caused by the scale mismatch between field plots and satellite pixels, this study selected the Zhangjiangkou National Mangrove Nature Reserve in Fujian Province as the study area and the mangrove distribution region of Fujian Province as the extrapolation area. A multiscale biomass estimation framework integrating field plots, unmanned aerial vehicles (UAVs), and satellite remote sensing was established. The results showed that (1) the optimal UAV-scale models achieved R2 values of 0.69 and 0.78 for aboveground biomass (AGB) and belowground biomass (BGB), respectively, with corresponding root mean square error (RMSE) values of 18.55 and 9.52 t·ha−1 and normalized root mean square error (nRMSE) values of 0.14 and 0.17 demonstrating reliable predictive performance; (2) after introducing UAV-derived bridging labels, the R2 of the AGB model increased from 0.24 to 0.64, while the RMSE decreased from 29.02 to 10.86 t·ha−1. Similarly, the R2 of the BGB model increased from 0.43 to 0.63, accompanied by a reduction in RMSE from 14.89 to 6.35 t·ha−1, demonstrating a substantial improvement in satellite-scale biomass estimation accuracy; (3) the total AGB and BGB of mangroves in Fujian Province were estimated at 58,768.60 t and 24,575.14 t, respectively, with high-biomass areas mainly distributed along the coastal regions of Zhangzhou and Quanzhou. Unlike conventional field-to-satellite extrapolation approaches, the proposed framework introduces UAV-derived biomass maps as intermediate bridging labels for pixel-level supervised learning, thereby establishing an effective link between field measurements and satellite observations. This strategy effectively reduces the scale mismatch between field and satellite data, significantly improves satellite-scale biomass estimation accuracy, and provides a transferable and scalable framework for regional mangrove biomass mapping and blue-carbon assessment.

Remote sensing image super-resolution aims to reconstruct high-resolution images from low-resolution observations and is important for image interpretation. Existing fixed-scale methods achieve good performance at predefined integer scales, but their dedicated upsampling modules limit their application to arbitrary-scale scenarios such as interactive GIS and multi-source image fusion. Continuous implicit methods provide scale flexibility but often exhibit spectral bias, resulting in over-smoothed textures and blurred object boundaries. To overcome these limitations, we propose an Arbitrary-scale Equivariant Resolution Operator (AERO) for remote sensing image super-resolution. AERO consists of three components. The Omnidirectional Feature Extractor enhances feature representation under orientation variations. The Wavelet–Arnold Residual Group models low- and high-frequency information in the wavelet domain to preserve textures and geographic boundaries. The Local Implicit Terrain Operator employs relative sub-pixel coordinates for continuous arbitrary-scale reconstruction. Experiments on AID, NWPU-RESISC45, UCMerced, and WHU-RS19 demonstrate that AERO achieves the best performance in the ×4 fixed-scale task. On WHU-RS19, AERO reaches a PSNR of 31.02 dB, exceeding FMSR by 0.64 dB. In rotational robustness experiments, the maximum PSNR fluctuation is reduced from 0.0181 dB to 0.0010 dB. The results show that AERO provides a practical approach for arbitrary-scale remote sensing image super-resolution.

Hyperspectral anomaly detection (HAD) identifies targets by spectral differences. However, large deep learning models overfit under the small-sample conditions typical of remote sensing, where anomalies are sparse and annotations costly. We propose a lightweight network on the GT-HAD transformer backbone for hyperspectral imagery. The design includes: (1) capacity reduction via multi-layer perceptron (MLP) ratio and embedding dimension optimization; (2) a Decoupled Projection Gating Fusion (DPGF) module enforcing spatial–spectral decoupling via dual-path projection and complementary regularization; (3) analysis of capacity–generalization tradeoffs across six airborne hyperspectral datasets. Experiments reveal an inverted-U relationship between capacity and generalization. The Mini configuration (121 K parameters) achieves a 53% parameter reduction. It yields AUC improvements on three datasets, maintains negligible performance loss (<0.1%) on two datasets, and exhibits a measurable decline on only one dataset, compared with the 255 K baseline. Notably, the baseline attains higher accuracy with only 25% training data on Pavia (+3.69% area under the receiver operating characteristic curve, AUC). This dataset-specific observation provides empirical evidence that capacity–data mismatch can induce severe overfitting in deep HAD models. Under a 25% training ratio, DPGF shows observable performance trends on spectrally homogeneous scenes, and zero initialization ensures identical performance to that of the baseline at the initial training stage, eliminating insertion risk during module deployment. However, the limited diversity of sensors and scene types constrains the generalizability of the observed trends. These results demonstrate that spatial–spectral decoupling with lightweight design suppresses overfitting in few-shot HAD, guiding compact models for onboard satellite and unmanned aerial vehicle (UAV) applications.

Abstract. Aerosol acidity significantly affects atmospheric chemistry and human health, yet its driving factors remain controversial. This study systematically examined a year-long characteristics of water-soluble inorganic ions in PM2.5 in the semi-arid Guanzhong Plain, Northwest China, and quantitatively analyzed its acidity and driving factors. The annual mean pH of PM2.5 was 3.8 ± 1.0 (winter > spring > summer > autumn). As pollution increased, aerosol pH shifted from the acidic range (2–5) on clean days to a near-neutral range (3–6) on polluted days. Sensitivity tests and driver analysis revealed that atmospheric temperature (22.3 %–33.8 %), NHx (gas NH3+NH4+, 11.4 %–44.9 %), and SO42- (8.5 %–10.8 %) were common key factors influencing pH across all seasons, with the percentages representing the quantitative contribution of each factor to the overall pH variation. Among these, temperature played a dominant role in seasonal variations, while NHx was the primary contributor during autumn and winter. Notably, Ca2+ emerged as a unique driver specific to spring, the relative standard deviation (RSD) was 8.8 %, indicating a substantial impact of Ca2+ variability on spring aerosol pH. Relative humidity (RH) exhibited a distinctive non-linear regulatory effect, i.e., aerosol pH initially decreases and subsequently increases with rising RH (inflection point occurred at 60 %–85 %). This phenomenon is primarily attributed to an abrupt change in liquid water content triggered by the deliquescence of hygroscopic components. Our results enhance the understanding of the contributions of various factors to aerosol pH.

Abstract. A substantial fraction of submicron particles originates from vehicle emissions in urban environments. This study investigated the chemical characteristics and sources of submicron organic aerosol (OA) at a traffic site in Helsinki, Finland, using four datasets collected in 2018–2024. Measurements were conducted using an Aerodyne Aerosol Mass Spectrometer, and source apportionment was performed using Positive Matrix Factorization. The results showed that vehicular traffic contributed to several types of OA. Hydrocarbon-like OA (HOA) typically peaked during morning traffic, whereas more oxygenated OA, referred to here as traffic-related OA (TrOA), also peaked in the morning but remained elevated for a longer duration. The mass spectra of TrOA resembled those of HOA and biomass burning OA, however, TrOA had distinct fractions of C2H4O2+ (at m/z 60), C2H5O2+ (at m/z 61) and C3H5O2+ (at m/z 73) in OA. The exact origin of TrOA remains uncertain, however, delayed morning peaks suggest that TrOA is processed in the atmosphere or emitted from modern vehicles, which typically operate later than heavy-duty vehicles. Semi-volatile OOA also appeared to be partially traffic-related, although due to its secondary nature, it was not directly linked to daily traffic patterns. This study highlights that traffic-associated OA encompasses both hydrocarbons and oxygenated POA and SOA. Relying solely on HOA to estimate traffic POA can result in a 50 % underestimation, as HOA and TrOA often have similar magnitudes. The characteristics of OA linked to vehicular emissions are likely to evolve in future as the vehicle fleet changes.

Anthropogenic heat is a central component of the urban energy balance. Until now it is primarily estimated by inventory methods or modelling approaches. In this study, we directly measured the anthropogenically emitted sensible heat from an air-conditioning system. Measurements of exhaust air temperature and volumetric flow rate were taken from 01 June 2025 to 31 August 2025 at a university building of Technische Universität Braunschweig, Germany. The anthropogenic sensible heat flux (Q H ) was calculated with these measurements and evaluated by descriptive statistics. Correlation and regression methods were used to investigate the relationship between Q H and meteorological variables. The sensible heat flux of air-conditioning ranged from 0 to 269 W m −2 building floor area (mean: 27 W m −2 ). Ambient air temperature and vapour pressure were found to be the strongest predictors. The regression slopes indicate that Q H from AC increased by approximately 3.6–4.5 W m −2 per 1 K increase in air temperature, and by 4.3–4.4 W m −2 per 1 hPa increase in vapour pressure. The observed co-dominance of air temperature and vapour pressure questions commonly used parameterisations of Q H based on air temperature alone, particularly for combined AC-ventilation systems. However, the moderate model performances of R 2 between 0.55 and 0.75 indicate additional influencing factors. This study provides the first direct measurements of anthropogenic heat from an air-conditioning system and offers reference values for inventory or modelling approaches.

Abstract Slow earthquakes, including episodic tremor and slip (ETS), are highly sensitive to small stress perturbations, yet their links to tectonic and non‐tectonic forcings remain unclear. Previous studies suggest that annual hydrological cycle dominates the tremor seasonal behavior in northern Cascadia. But over a longer time range, the tremor seasonal pattern seems significantly changed. Here, we systematically examine major stress perturbations toward influencing tremor and slow slip event (SSE) activities in Cascadia and identify two distinct patterns: an incongruent tremor seasonal variation that cannot be explained by hydrological forcing and a long‐term change in the locations where ETS events nucleate and their propagation directions. The tremor seasonality is better explained by stress variation due to SSEs, while the long‐term ETS evolution likely reflects spatiotemporal stress redistribution on subduction fault interface through stress feedback of past major ETS events and potential stress balancing along fault strike.

Introduction Understanding how coastal communities perceive environmental change is critical for designing effective and locally supported marine management strategies. This study examined perceptions of environmental changes over the past 10 years and expectations for the next 10 years among coastal residents of the Cù Lao Chàm–Hội An Biosphere Reserve in central Vietnam. Methods Face-to-face surveys were conducted with 252 adult residents from mainland and island communities. Participants assessed past and expected future changes in different aspects of the marine environment. Principal component analysis was used to identify the main dimensions underlying these perceptions, followed by confirmatory factor analysis to assess model fit. One-sample t-tests and repeated-measures analyses of variance were conducted to examine perceived changes over time and differences between mainland and island communities. Results Four dimensions of perceived marine environmental change were identified: marine resources, seawater quality, aquaculture, and marine habitats. The four-factor model showed acceptable fit. Respondents perceived significant declines in marine resources and aquaculture over the previous decade and expected these declines to continue. Seawater quality was perceived to have deteriorated slightly, with no clear improvement expected in the future. In contrast, marine habitats were perceived to have improved and were expected to continue improving. Perceptions differed significantly between mainland and island residents, particularly regarding seawater quality and marine habitats. Discussion Community perceptions of marine environmental change were structured, place-specific, and broadly consistent across past and future timeframes. Differences between mainland and island communities suggest that spatial context, livelihood dependence, and direct interactions with marine environments shape how environmental change is experienced and anticipated. Integrating community perceptions with ecological evidence may improve alignment between conservation measures and local expectations, thereby supporting more adaptive, effective, and socially accepted marine governance.

Climate change poses a significant threat to the sustainability of smallholder livestock systems in Sub-Saharan Africa (SSA), where livelihoods depend heavily on climate-sensitive natural resources. Although considerable research has examined climate impacts and adaptation strategies, evidence on priority research areas required to strengthen resilience remains fragmented. This systematic review synthesized existing evidence on climate adaptation strategies in smallholder livestock systems and identified key knowledge gaps and future research priorities in SSA. A systematic literature search was conducted following PRISMA 2020 guidelines using Scopus and Web of Science databases, complemented by manual searches. A thematic synthesis approach was employed to analyse evidence from 64 studies published between 2000 and 2025. Climate change affects smallholder livestock systems through feed and water scarcity, rangeland degradation, increased disease prevalence, and declining livestock productivity. Most adaptation strategies remain incremental. These include the use of indigenous breeds, feed supplementation, livestock mobility, water harvesting, and mixed crop-livestock systems. Their effectiveness is often constrained by weak institutional support, limited access to extension services and finance, and inadequate infrastructure. The review also identified important knowledge gaps in climate-resilient livestock genetics, sustainable feed systems, climate-sensitive disease surveillance, socioeconomic resilience, and digital and policy innovations. Addressing these priorities is essential for developing transformative, climate-resilient, and sustainable smallholder livestock systems in SSA.

This study proposes a prediction method for the identification of areas inside and around cities, that—based on their morphometric and built-up conditions—are prone to develop as hot spots in the highly probable case of cities development. To achieve this objective, hot and cold spots of air temperature distribution for the six largest cities from north-eastern Romania were firstly identified using data from mobile measurements, performed during the warm season. From May to September 2022, 64 mobile measurements were made following a standardized observation plan. The measurements were carried out under calm and stable atmospheric conditions, with clear or partly cloudy skies, before sunrise and immediately after sunset in order to ensure representativeness. Global Moran’s Index was used to assess spatial autocorrelation, while Getis-Ord Gi* was used for hot and cold spot identification. The results describe accurately the built-up ratio and landform morphology conditions of the areas that are warmer (hot spots)/colder (cold spots) than their surroundings. In brief, it can be observed that the occurrence of hot spots is mostly controlled by high imperviousness density values (>45%), while cold spots are shaped mainly by local natural conditions that are favorable for the accumulation of cold air below the thermal inversion band. Secondly, based on the relationship of altitude and built-up ratio with hot spots occurrence, we developed a prediction model of the urban areas that are prone to sustain or evolve into hot spots. The analysis results are meant to serve cities stakeholders involved in the mitigation of the urban heat island effects, helping them to identify the regions that are in risk of becoming excessively warm during future summers.

Abstract Aerosol phase state governs condensed-phase diffusion and heterogeneous reactivity, yet viscosity constraints for ambient urban PM2.5 remain scarce. Here, we quantified relative humidity (RH)-dependent viscosities of urban PM2.5 collected in Ansan, South Korea, during summer 2024 using a poke-and-flow technique coupled with fluid-dynamics simulations, yielding 6.2 × 104 to 1.3 × 107 Pa s at RH ∼20–40% and exceeding ∼108 Pa s at RH < ∼10%. Integrating the Ansan data set with previously reported PM2.5 viscosities from Seoul and Beijing, a unified temperature–RH parametrization was derived using the Vogel–Tammann–Fulcher framework and applied via a machine-learning surrogate to hourly meteorological data from 14 global megacities over September 2023–August 2024. Predicted PM2.5 viscosity was systematically lower at night than during the day across all cities, driven by RH-induced plasticization that outweighed nighttime cooling, although these predictions were based on 24 h PM2.5 composition and therefore did not include time-resolved changes in aerosol chemical composition. Incorporating viscosity-dependent N2O5 diffusivity into a resistor-model framework yielded mean nighttime N2O5 uptake coefficients of approximately 0.01–0.07, up to approximately 1 order of magnitude below the conventional liquid-particle assumption of 0.1, though substantial uncertainty remains from propagated viscosity measurement errors and parametrization assumptions. These results suggest that viscosity measurements from field-collected PM2.5 can provide improved estimates of diurnal aerosol-phase conditions relevant to nocturnal N2O5 reactivity and nitrate formation in urban atmospheres.

Optical–elevation data fusion is widely used in aerial remote sensing semantic segmentation, as optical imagery provides rich spectral and textural information, while DSM or DEM data offer complementary elevation-related structural cues. However, effective fusion remains challenging because optical and elevation representations may exhibit cross-modal structural inconsistency, frequency–spatial response imbalance, and decoder-stage structural attenuation. To address these challenges, we propose GCF-Net, a stage-aligned optical–elevation fusion network that matches different cross-modal processing objectives to the evolving representation states of the encoder–decoder pipeline. A Structure-Guided Cross-Modal Correction Module first performs structure-conditioned correction of modality-specific features before fusion. A Frequency–Spatial Cross-Modal Fusion Module then constructs joint representations through bounded cross-modal magnitude conditioning, frequency-to-spatial reconstruction, and spatial recalibration. During decoding, a Geometry-Aware Cross-Scale Refinement Module reintroduces elevation-derived structural guidance into multiscale fused features. Experiments on ISPRS Vaihingen, ISPRS Potsdam, and MMHunan yield mIoU scores of 72.37%, 75.38%, and 52.23%, respectively, achieving the highest mIoU among the evaluated unimodal, multimodal, and SAM-based methods under the unified protocol. Ablation and replacement experiments verify the complementary roles of the three stage-specific components, while sensitivity, elevation perturbation, and complexity analyses indicate architectural flexibility, tolerance to moderate elevation degradation, and a balanced accuracy–efficiency trade-off.

Abstract High-altitude saline lakes on the Tibetan Plateau host some of the world’s most arsenic-rich waters, yet controls on arsenic (As) transformation remain poorly resolved. By integrating depth-resolved geochemistry with metagenomic profiling in Lake Tosen, we identified a redox framework for As cycling that is not fully explained by the Fe-centric model derived from groundwater and freshwater. In summer, oxic-suboxic-anoxic stratification supported three distinct pathways: (i) oxic waters exhibited a microbial detoxification-oxidation loop sustaining As(V) dominance; (ii) suboxic waters were associated with nitrate-supported As(III) oxidation linked to denitrifying taxa; and (iii) anoxic sulfidic waters showed sulfate-reduction-related As transformation, with sulfur cycling likely modifying the fate and mobility of released As under sulfidic conditions. Meanwhile, canonical Fe(III)-respiration markers were not prominent, and saturation indices indicated stable goethite/hematite under in situ conditions. Together, these findings indicate that, under the observed stratified conditions in Lake Tosen, nitrate- and sulfate-associated pathways were more closely linked to arsenic transformation than canonical Fe(III)-respiration signals. This nonclassical coupling may become increasingly relevant as stratification, nitrate loading, and salinization intensify, with implications for As risk prediction in endorheic basins under changing climates.

Abstract. This study extends previous research on CMIP5 models to investigate the reproducibility of climate models within the Coupled Model Intercomparison Project (CMIP). It evaluates the accessibility to the source code of the CMIP models through all their phases, emphasizing the need for public repositories to ensure transparency regarding model input, output, and usage rights, along with an analysis of licenses for compliance with scientific standards. A central focus of the research is the assessment of code quality against best practices. In addition, the study examines the historical evolution of computational and code quality across various phases of CMIP, highlighting progress and improving traceability to support scientific reproducibility. We provide valuable insights for future research, proposing solutions and tools designed to improve replicability and enhance project lifecycles that are applicable not only to CMIP but also to broader scientific contexts.

Wind-induced tree motion in forests reflects the interaction between atmospheric forcing and tree-specific mechanical properties, yet the extent to which stand-scale response is governed by coherent wind forcing or by individual tree characteristics remains unclear under field conditions. Here, wind-induced response of 28 neighboring trees in a planted Scots pine forest stand was analyzed using singular value decomposition and maximum covariance analysis to identify coupled wind–tree components and to partition tree response into collective and individual shares. The results reveal that tree response is dominated by a single wind-driven coupled component, onto which the response of all trees projects to a large extent. This enables a consistent decomposition into a collective response, representing the shared stand-scale dynamics, and an individual contribution capturing tree-specific deviations. Collective participation is systematically associated with alignment to the dominant wind-driven component, whereas individual contributions exhibit greater variability and do not show a consistent dependence on intrinsic dynamic properties or local structural factors such as neighborhood configuration. Partial correlation analysis indicates that neither fundamental sway frequency nor crowding index explains additional variation in wind–tree coupling beyond that associated with participation in the dominant collective response. These findings demonstrate that, in the structurally homogeneous stand, tree response is primarily governed by stand-scale wind loading, expressed by a single wind-driven coupled mode accounting for approximately 82 % of the variance in the aligned tree response field. Individual tree properties primarily modulate deviations from this dominant collective response. This provides a reduced-order, process-based perspective on wind–tree interaction in the planted Scots pine stand and highlights the central role of collective dynamics under natural wind conditions.

In karst regions, sugarcane mapping faces challenges from fragmented fields, undulating terrain, spectral confusion, and persistent cloud cover, which limit traditional optical remote sensing. To address these issues, we propose a fine-scale extraction framework that integrates Sentinel-2 optical and Sentinel-1 synthetic aperture radar (SAR) imagery through image-level fusion, and introduces a UNet-DFH network with a Multi-Scale Edge Fusion (MSEF) module and an Attention-Deformable Fusion Module (ADFM). This study makes three core contributions: (1) we construct a dedicated optical–SAR collaborative sugarcane extraction dataset for typical karst regions, alleviating the scarcity of multimodal labeled samples; (2) we propose the UNet-DFH network, where MSEF enhances boundary preservation and topological detail in shallow decoding stages, while ADFM improves robustness to geometric deformation and local misalignment in deep semantic stages; (3) we demonstrate that the joint mechanism of edge-preserving filtering and deformable adaptation yields a synergistic effect in addressing the precision–recall trade-off. Experiments in a typical karst area of Guangxi, China, demonstrate that optical–SAR fusion achieves an IoU of 80.08% and an OA of 92.09% during the sugar accumulation and maturity stage. During the more challenging tillering stage, UNet-DFH maintains relatively stable performance under optical-only conditions, with an IoU of 72.98%, Recall of 82.78%, and OA of 92.12%. Moreover, optical–SAR fusion improves Recall by 5.5 percentage points over optical-only inputs (from 83.54% to 89.04%), while Precision exhibits a moderate decrease from 91.89% to 88.84%, reflecting the expected trade-off associated with speckle noise. These results confirm the complementary value of multimodal data and the effectiveness of the proposed modules in preserving fragmented plot boundaries and improving segmentation performance in complex karst terrain. The framework offers a promising approach for high-precision crop mapping in the studied karst agricultural landscape.