Soil heavy metals (HMs) pose a serious risk to the environment owing to their inherent accumulative and non-degradable characteristics. Artificial intelligence (AI) models demonstrated as powerful data-driven tools to mimic the existence of HM in soil and has so far contributed remarkably to the simulation of their environmental aspects. In this perspective, several concerns on the development of AI models for soil HM modeling were discussed. Data quality, optimal feature selections, models’ interpretability, transferability, and generalization have been debated to show how these limitations can cause to misleading or non-causal patterns, thereby directly impeding the rapid knowledge discovery phase from intelligent data-driven methods. As demonstrated in the graphical abstract, soil heavy metals (HMs) represent a critical environmental threat on earth system due to their non-degradable nature and tendency to accumulate over time. Artificial Intelligence (AI) models have emerged as a powerful tool for quantifying and simulating HM presence in soil, offering a sophisticated way to mimic complex environmental behaviors. However, there are several concerns in the literature required serious attention by developers including data quality, feature selection, and model interpretability. In addition, achieving better transferability and generalization is considered vital for reliable soil HM quantification across diverse geographic regions. In this communication article, those concerns were debated and discussed to provide better understanding of those concerns where an advancement proposed to facilitate the significant knowledge discovery, enabling more precise environmental risk assessments. By refining these models, researchers/developers/designers can better build a computer aid predictive models for HM distribution, ultimately supporting more effective remediation strategies to protect global food security and ecosystem health. Key challenges and unresolved issues in AI models for soil HM simulation were critically synthesized. Data quality, optimal feature selections, models’ interpretability were discussed for better AI models development. Model transferability and generalization were debated for better soil heavy metals quantification. We propose possible frontiers for unresolved challenges in soil heavy metals prediction under next-generation research era. We envisage that our critical contributions will lead to greater adoption of AI models for practical purposes in soil health applications.
This study examines the infection progression of tuberculosis (TB) and SARS-CoV-2 co-infection using a mathematical modeling framework combined with Artificial Neural Networks (ANNs). A system of ordinary differential equations is developed to characterize the interactions between the two diseases transmission. Numerical solutions generated through the 4^th -order Runge-Kutta (RK4) technique in MATLAB are employed to train the ANN model, with the dataset split into 81 The graphical abstract presents a comprehensive model for analyzing the co-infection dynamics of (TB) and SARS-Cov-2 through a mathematical model integrated with Artificial Neural Networks (ANNs). The main compartments are: susceptible 𝕊 , exposed classes 𝔼_1 and 𝔼_2 , infected classes 𝕀_1 and 𝕀_2 , co-infected class ℂ , and recovered population ℝ . A system of differential equations governs the transitions between these compartments. Numerical solutions are obtained using the RK4 approach in MATLAB, which are subsequently utilized to train the ANN model. The dataset was systematically partitioned into training (81
Accurate solar resource forecasting is essential for renewable energy expansion in climate-sensitive coastal regions, where atmospheric variability affects power generation and grid stability. In Bangladesh, the southeastern coastal zone experiences strong seasonal fluctuations driven by Bay of Bengal monsoon dynamics, creating challenges for reliable solar energy planning and operational forecasting. Despite advances in artificial intelligence, limited studies have integrated spatial climate interactions, long-term environmental observations, and explainable forecasting approaches for coastal solar resource assessment. This study develops an Earth System Intelligence framework for solar irradiance forecasting and energy potential mapping across the southeastern coastal region of Bangladesh. A harmonized 25-year dataset (2000–2025) was compiled from the Bangladesh Meteorological Department and NASA POWER, covering 12 coastal meteorological stations. Data were processed using the APS7C framework, incorporating nighttime masking, Butterworth filtering, normalization, and cyclical temporal encoding. Four forecasting architectures were evaluated: Spatio-Temporal Graph Attention Transformer (STGAT), Informer, CNN-LSTM-Attention, and a Hierarchical Ensemble Stack. The ensemble framework integrates Informer, CrossFormer, STGAT, and GRU base learners through a LightGBM meta-learner to capture spatial dependencies and temporal climate variability. The Hierarchical Ensemble Stack achieved the best predictive performance, with a correlation coefficient of 0.931 and a normalized standard deviation of 0.992. The Informer model maintained root mean square error values below 32 W/m² across 24–72 h forecasting horizons. The Hierarchical Ensemble Stack achieved the highest accuracy (correlation coefficient = 0.931; normalized standard deviation = 0.992), improving correlation by 8.4 This graphical abstract presents an Earth System Intelligence framework developed to support solar resource assessment and coastal energy planning in southeastern Bangladesh. The visual summary is organized into three interconnected components: Data Analysis, Models Learning Framework, and Results Applications. The first component highlights the integration of a 25-year coastal solar irradiance dataset (2000–2025) derived from the Bangladesh Meteorological Department (BMD) and NASA POWER across 12 meteorological stations. Data preparation was performed using the APS7C framework, which includes filtering, normalization, quality control, and feature engineering to enhance data reliability and model performance. The second component summarizes the forecasting architectures evaluated in the study, including Spatio-Temporal Graph Attention Networks (STGAT), Informer, CNN-LSTM-Attention, and a hierarchical ensemble framework with a LightGBM meta-learner. These models were designed to capture both temporal variability and spatial interactions within the coastal climate system. The final component presents the major outcomes and practical significance of the research. The ensemble framework achieved the highest predictive performance with a correlation coefficient of 0.931, while maintaining RMSE values below 32 W/m² across forecasting horizons. SHAP analysis identified the most influential predictors of solar irradiance variability, and spatial learning mechanisms revealed inter-station dependencies. The resulting framework provides a foundation for resource mapping, energy planning, grid integration, and data-driven renewable energy management in coastal environments. Developed an Earth system intelligence framework using 25 years of coastal climate observations. Integrated STGAT, Informer, CNN-LSTM-Attention, and ensemble learning for GHI forecasting. Hierarchical Ensemble Stack achieved R = 0.931 and normalized standard deviation of 0.992. SHAP analysis identified short-term irradiance persistence as the dominant predictor. Forecast outputs support energy planning, solar mapping, and smart-grid decision support.
Effective watershed management in semi-arid regions requires a precise understanding of how land use change (LUC) and climate change (CC) will jointly alter hydrological cycles. This study addresses a critical gap in assessing their relative impacts by evaluating the individual and combined effects of future LUC and CC on key hydrological variables in the Siminehrud River Basin, Iran, a region vulnerable to water scarcity. The Soil and Water Assessment Tool (SWAT) model was calibrated and validated for streamflow simulation. Future LUC for 2040 was projected using the Dyna-CLUE model under three distinct scenarios: sustainable development (MTSS), rapid expansion (MSSS), and business-as-usual (MBAU). Climate projections from five General Circulation Models under RCP4.5 were statistically downscaled using LARS-WG. Results indicated a marginal increase in annual precipitation (0.9 This graphical abstract provides a rapid, visually compelling summary of a comprehensive hydrological impact assessment study in the Siminehrud River Basin, Iran. It illustrates the principal methodological steps, beginning with the application of IPCC Representative Concentration Pathways (RCPs). Future climate projections from multiple Global Climate Models (under RCP4.5) and three distinct land-use scenarios for 2040, Sustainable Development (MTSS), Rapid Expansion (MSSS), and Business-as-Usual (MBAU), form the core inputs. These scenarios were developed using the Dyna-CLUE model, the results of which are visualized as classified land-use maps highlighting changes in Built-up areas, Irrigated cultivation, Rainfed agriculture, and Rangelands. The central flowchart depicts how these climate and land-use change projections are integrated into the SWAT (Soil and Water Assessment Tool) hydrological model for simulation. The right section presents a key comparative result, showing average monthly evapotranspiration (ET) and outflow (streamflow) variations for the combined scenarios against the baseline. The chart visually encapsulates the study's main finding: while future precipitation shows a slight increase, rising temperatures and land-use change drive a critical hydrological shift. The simulations project an overall decrease in annual streamflow alongside significant increases in ET under most scenarios. Crucially, the analysis revealed that land-use change was the dominant driver of hydrological alteration, surpassing the impact of climate change alone. This graphical abstract efficiently conveys the study's core warning: sustainable watershed management must explicitly address the trade-off between land-use development strategies and their unintended consequences on water scarcity, as visualized through the divergent pathways of streamflow and ET. Land-use change dominates future hydrological alteration, surpassing climate change impacts. Land-use intensification raises evapotranspiration (ET) while reducing streamflow, worsening scarcity. Future streamflow declines up to -3.1 to -4.5
Accurate, high-resolution local climate projections are essential for climate adaptation and impact assessment. Most statistical downscaling approaches rely on regression-based models that produce single deterministic estimates, thereby limiting their ability to capture the inherent variability and multivariate dependencies of the climate system. To address this limitation, ClimDiT is introduced, a generative multivariate statistical downscaling framework based on latent diffusion transformer models. Thus, instead of a single expected result, it estimates the full conditional distribution of high-resolution climate fields. The proposed system operates within a compressed, autoencoded latent space, enabling computationally efficient sampling while implicitly capturing inter-variable dependencies. In this work, ClimDiT simultaneously downscales daily maximum temperature, minimum temperature, and accumulated precipitation to approximately 5 km resolution over the Iberian Peninsula. It is trained using ERA5 predictors and ROCIO-IBEB high-resolution targets and its performance is evaluated using metrics aligned with the VALUE framework. ClimDiT attains deterministic accuracy comparable to, or moderately superior to, state-of-the-art regression networks, while improving spatial coherence. Although specialised stochastic baselines, which suffer from high deterministic errors, retain an edge in univariate and multivariate probabilistic scores, a ClimDiT ensemble generated from 50 samples demonstrates reasonable probabilistic skill and joint calibration of the three target variables (evaluated with CRPS, Brier, Energy and Variogram scores). Overall, ClimDiT demonstrates that latent diffusion models enable a generative pathway toward coherent, probabilistic, and physically consistent regional climate downscaling, advancing beyond the limitations of traditional regression-based AI systems. This graphical abstract provides a concise visual overview of ClimDiT: A Generative Latent Diffusion Transformer Framework for Multivariate Climate Downscaling. It summarizes the study objectives, methods, and main findings. At the top, the Data section illustrates the statistical downscaling approach, by linking coarse-resolution ERA5 predictors with high-resolution ROCIO-IBEB climate targets over the Iberian Peninsula. It highlights the three downscaled variables—maximum temperature, minimum temperature, and precipitation—as well as the corresponding training and testing periods. The Methods panel presents the core innovation: the ClimDiT model. It combines an autoencoder, which compresses climate fields into a latent space, with a latent diffusion transformer that progressively reconstructs high-resolution fields from Gaussian noise, conditioned on large-scale predictors. This reflects the proposed use of generative models, rather than the widely used regression-based AI approaches, as they allow the generation of multiple outputs for the same synoptic state. A benchmark comparison box displays the univariate state-of-the-art (SOTA) methods (DeepESD and U-Net) used as references, along with the tested loss functions (MSE, ASYM, and STO). STO refers to stochastic models, which use a Bernoulli–Gamma distribution for precipitation and a Gaussian distribution for temperature, trained using the negative log-likelihood loss. Additionally, a deterministic multivariate baseline (DeepESD-MULT, trained with MSE) is included to directly evaluate the impact of joint variable modeling within a traditional regression framework. The Results section visualizes performance across spectral, deterministic, and probabilistic metrics (RAPSD, MAE, CRPS, Variogram Score), emphasizing ClimDiT superior spatial coherence and balanced probabilistic skill. The Highlights box succinctly reinforces these contributions: a generative, latent-space-based approach that enables coherent multivariate downscaling.
Carbon dioxide (CO₂) fluxes in terrestrial ecosystems, particularly Net Ecosystem Exchange (NEE), Gross Primary Productivity (GPP), and Ecosystem Respiration (RE), are fundamental indicators of biosphere–atmosphere carbon exchange and play a central role in regulating the global carbon cycle. However, their accurate quantification in subtropical grasslands remains challenging due to complex nonlinear interactions among climatic drivers and increasing environmental pressures. This study addresses this limitation by applying advanced machine learning models to predict CO₂ flux components in the Southern Brazilian Pampa, an ecologically important yet underrepresented biome. Meteorological and flux data were obtained from eddy covariance towers at Aceguá (2018–2022) and Santa Maria (2014–2023) at half-hourly resolution. Six models (multiple linear regression (MLR), artificial neural network (ANN), random forest (RF), extreme gradient boosting (XGB), support vector machine (SVM), and decision tree (DT)) were evaluated, trained and tested using solar radiation, air temperature, relative humidity, wind speed, and vapour pressure deficit as predictors. Model performance was evaluated using correlation heatmaps, statistical metrics, Taylor diagrams, violin plots, and multi-temporal analyses. Results from the heatmap showed very strong physical relationship, with NEE negatively correlated with solar radiation (R = − 0.93 to − 0.96), while GPP exhibited strong positive correlations (R > 0.94), and RE was closely linked to air temperature (R = 0.87–0.91). Ensemble models (RF and XGB) outperformed others, particularly in Santa Maria, achieving high predictive accuracy for NEE (R² = 0.98 and 0.98), GPP (R² = 0.99 and 0.98), and RE (R² = 0.95 and 0.94). Taylor diagrams indicated strong agreement between predicted and observed RE (σ = 7.3 µmol m⁻² s⁻¹; R > 0.93), supported by close median values (actual RE: 7.32; XGB: 7.34; RF: 7.29). In Aceguá, RF and ANN provided the best RE predictions (test R² = 0.86), though with slightly higher variability. Finally, the ensemble models demonstrated strong predictive performance under the evaluated within-site chronological validation design and study limitations. This shows the potential usefulness of machine learning approaches for site-level carbon flux assessment in subtropical grassland ecosystems. This graphical abstract provides a concise and visually engaging summary of the study on optimizing CO₂ flux quantification in Southern Brazilian Pampa agroecosystems using machine learning techniques. It integrates key elements of the research workflow and findings into a single visual narrative. The upper section presents the study sites (Aceguá and Santa Maria) alongside the primary meteorological drivers, including solar radiation, temperature, relative humidity, wind speed, and vapour pressure deficit. The central panel highlights the application of multiple machine learning models, such as Random Forest, XGBoost, Artificial Neural Networks, Support Vector Machines, Multiple Linear Regression, and Decision Trees for predicting carbon flux components: Net Ecosystem Exchange (NEE), Gross Primary Productivity (GPP), and Ecosystem Respiration (RE). Notably, ensemble approaches achieved the highest predictive accuracy, with coefficients of determination reaching R² ≈ 0.98 for NEE, R² ≈ 0.99 for GPP, and R² ≈ 0.95 for RE. The correlation heatmaps reveal strong relationships between variables, such as GPP exhibiting correlations up to 0.96 with radiation and NEE showing inverse correlations of about − 0.96 with productivity-related variables. The lower figures present Taylor diagrams, where model performance clusters close to the reference point, indicating low standard deviation differences and high correlation (R > 0.90) during both training and testing phases. Overall, this graphical abstract serves as a quick, visually appealing summary of the research, enabling rapid understanding of the study’s objectives, methods, and major outcomes without requiring a full reading of the manuscript. Ensemble models (RF, XGB) achieved top accuracy with R² up to 0.99 for GPP, 0.98 for NEE, and 0.95 for RE across sites. Strong correlations found: GPP–radiation (r ≈ 0.94–0.96), NEE–radiation (r ≈ − 0.93 to − 0.96), RE–temperature (r ≈ 0.87–0.91). Santa Maria models showed superior stability with R > 0.98 and low RMSE ( 0.26–0.45 µmol m⁻² s⁻¹) for GPP and RE predictions. Test results confirmed strong predictive skills of RF and XGB models, having maintained R² ≈ 0.95 for NEE and GPP, with low prediction errors across datasets. Temporal analysis showed models captured seasonal CO₂ flux dynamics accurately, maintaining consistency across hourly to monthly scales.
Natural disasters such as floods, cyclones, and earthquakes cause damage to life and property. India’s climate is prone to various natural disasters. Disaster situation monitoring is essential for efficient disaster management. Social media is one of the quickest ways to share real-time information, making it crucial for analysing and monitoring natural disasters. Relying solely on information found online is inadequate due to its lack of credibility and the risk of misinformation. The main challenge lies in extracting disaster-related information from text data. Text feeds can provide disaster type, location, and urgency sentiment. With the location and disaster type extracted, relevant authorities can make informed decisions for relief and rescue missions. Instead of conventional three-class sentiment analysis, the proposed system performs binary urgency classification, in which the “negative” class is treated as a proxy for urgent disaster information. The “positive” and “neutral” classes are combined into a single “not negative” class, representing non-urgent information related to preparedness, recovery, and general updates. The proposed solution integrates transformer-based Natural Language Processing (NLP) models, news article and weather station data retrieval and performs automated information extraction. The proposed operational natural disaster monitoring system also validates its performance through case studies against official meteorological and disaster-reported data. Models such as Bidirectional Encoder Representations from Transformers (BERT) and Robustly Optimized BERT Pretraining Approach (RoBERTa) are fine-tuned on custom datasets created from historical news articles and tweets for disaster news extraction in this study. The BERT model is used to apply a relevance filter over text (binary classification) and achieves an accuracy of 91.52 https://drive.google.com/drive/folders/1tFHTHYKLntKaWyjz-2yIeRVSzuhsD2Z4 . The AI-based natural disaster information monitoring system is illustrated in a graphical abstract. It consists of four modules. The first module focuses on model development. The disaster data are collected from social media tweets and historical news articles. The data are manually labelled to create two datasets, one for urgency sentiment analysis and another for relevance filtering. For urgency analysis, data are classified into two classes: “negative” and “not negative.” The neutral and positive are combined into a single class (“not negative”) to improve task efficiency and accuracy. For relevance filtering, data with natural disaster keywords were labelled as relevant or irrelevant. The BERT model is fine-tuned for the relevance filtering task, while RoBERTa is fine-tuned for binary urgency sentiment analysis. The second module fetches news from Really Simple Syndication (RSS) feeds of Indian news channels, pre-processes the data to extract relevant features, and then stores the data in a relational database. The news feed information extracted by the second module is further validated by the third module. The third module is designed to retrieve data published by the India Meteorological Department’s weather stations. The dataset includes meteorological warnings and data on rainfall, wind speed, and temperature over the study area, all stored in a relational database (PostgreSQL). The disaster information processed by the deep learning model is superimposed onto the real-time IMD data for validation. The final module is the development of a web-based spatial information system for data visualisation and analysis. It features a spatial dashboard that enables users to view and analyse information collected from the second and third modules for disaster situation assessment and monitoring. Urgency sentiment analysis plays a significant role in assessing disaster situations. It provides a detailed and synoptic view of the disaster situation across India in near real-time by analysing data from over 220 RSS feed links. This timely disaster information supports the planning of satellite data acquisition for flood mapping and damage assessment. It enables efficient utilisation of Indian Remote Sensing (IRS) satellites through adaptive sensor tasking over affected areas identified by the AI-based natural disaster information monitoring system. Introduces custom-labelled datasets for text relevance filtering, urgency sentiment analysis, and a comprehensive table of local and national-level news sources. Proposes a transformer model-based pipeline to reduce noisy data and perform information extraction. Demonstrates a fifteen-day case study validating results from the AI system against official rainfall forecast products in the Delhi and Gujarat region. Integrates the AI pipeline with a web GIS for disaster event monitoring and management. Supports integration with advanced Indian Remote Sensing Satellite datasets, optimising satellite coverage, and minimising data acquisition costs.
Extreme heat poses escalating risks to human health, labor productivity, and infrastructure resilience in hyper-arid regions. As climate change intensifies the frequency and severity of heatwaves, operational early warning systems with meaningful lead time are critical for strengthening climate resilience and supporting adaptation to extreme heat. In this study, we developed a hybrid machine learning (ML) framework for operational 24-hour-ahead forecasting of the Universal Thermal Climate Index (UTCI) in Iraq, one of the world’s most heat-vulnerable regions. The framework integrates a Twin Support Vector Machine (TSVM) optimized using a hybrid Whale Optimization Algorithm–Salp Swarm Algorithm (HWOA-SSA) to effectively capture the nonlinear and highly dynamic relationships among air temperature (AT), solar radiation (SR), relative humidity (RH), and wind speed (WS). The developed TSVM was validated against Support Vector Machine (SVM), Relevance Vector Machine (RVM), and Random Forest (RF) models. Using hourly ERA5 reanalysis data (1981–2020), the model was trained to generate high-resolution UTCI forecasts with a 24-hour lead time, and its performance was evaluated using a chronological calibration-validation split, in which the earlier portion of the time series (1981–2008) was used for model calibration and the later period (2009–2020) was reserved exclusively for validation. The optimized TSVM demonstrated strong predictive performance on temporally independent validation data, achieving a Kling–Gupta Efficiency (KGE) of 0.983, a correlation coefficient (R²) of 0.991, and a normalized RMSE of 0.206. The framework accurately reproduced diurnal amplification patterns and the spatial heterogeneity of extreme heat stress across Iraq, including regions that frequently exceed very strong heat stress thresholds during summer afternoons. By enabling reliable lead-time forecasting of hourly heat stress, the proposed framework offers a promising hybrid machine-learning approach for future heat early-warning applications and may be adapted to other hyper-arid and heat-prone regions through local calibration and validation. As shown in the graphical abstract, a hybrid data-intelligence framework was developed to predict thermal comfort under a changing climate. The study begins by identifying the growing impact of climate change on thermal stress, recognizing the global increase in heat exposure and its consequences for human health and livability. To build a robust predictive foundation, high-resolution meteorological data from the ERA5 dataset are utilized, providing reliable atmospheric variables for model development. Machine learning (ML) models are then developed to forecast the Universal Thermal Climate Index (UTCI), enabling accurate prediction of thermal comfort conditions. The methodology incorporates an innovative hybrid metaheuristic optimization approach—HWOA-SSA (a combination of Whale Optimization Algorithm and Salp Swarm Algorithm)—to enhance parameter tuning and improve predictive efficiency. This hybrid strategy accelerates convergence, avoids local minima, and increases overall model robustness. A focused regional case study is conducted in Iraq, a climate-vulnerable region characterized by extreme heat stress. This regional application allows for detailed investigation of local meteorological drivers and model adaptability under severe thermal conditions. Model performance is rigorously evaluated using statistical accuracy metrics to assess reliability and generalizability. Correlation analysis is performed to identify significant relationships between UTCI and key meteorological variables, clarifying the relative influence of temperature, humidity, wind speed, and radiation on thermal comfort. The optimized hybrid framework demonstrates measurable performance improvements compared to conventional models. Finally, the study highlights the model’s global applicability, suggesting its potential effectiveness for other regions experiencing rising thermal stress. Overall, the framework integrates climate science, advanced data analytics, and optimization techniques to support sustainable climate adaptation and heat risk management strategies. A hybrid TSVM-HWOA-SSA framework was developed for forecasting of hourly UTCI over Iraq. ERA5 hourly meteorological variables were used to construct lead-time forecasting datasets. Model skill was evaluated using a chronological split to ensure temporally independent testing. The optimized TSVM outperformed other ML models in forecasting UTCI. The framework successfully forecasted the diurnal variability and spatial heterogeneity of extreme heat stress across Iraq.
Rapid urbanisation and land use transformation pose significant threats to long-term ecological sustainability, particularly in the eco-environmental conditions of the fastest-growing metropolitan areas in developing nations. The Kolkata metropolitan area has posed a critical ecological health quality over the past three decades due to fast-growing urbanisation and human activities. The study aims to assess the impact of urbanization, land use land cover (LULC) dynamics on the ecological health quality of the Kolkata Metropolitan Area (KMA). Multidate Landsat5 TM for 1990, 2000, and 2010, and Landsat8 OLI for 2020 were utilised in this study. First, four Machine Learning (ML) algorithms were tested, and the best algorithm was used to generate the LULC maps from 1990 to 2020. Land use transition and the estimation of the 2030 projection of LULC were carried out using Cellular Automata-Artificial Neural Network. Second, multidate Remote Sensing-based Ecological Health Index (RSEHI) was developed by integrating four spectral indices and the Analytic Hierarchy Process to evaluate the spatiotemporal variability of the quality of ecological health. Lastly, multivariate fractional regression was used to predict the RESHI and it establish comprehensive assessment of how urbanization and LULC changes directly influence the ecological health quality of the KMA. The results indicate a consistent decline in ecological health due to rapid urban expansion. The built-up area experienced a significant increase, strongly and negatively correlated (r = -0.857) with RSEHI, whereas vegetation cover shows perfect and positive correlation (r = 0.994), and is identified as the most promising factor, which is rapidly declining over time. The city centre resulting a zone with low ecological health quality, expected to expand to 1032.27 km² (57.51 This graphical abstract illustrated the integrated methodology adopted for spatial assessment of urbanization and land use land cover dynamics on urban ecological health in the Kolkata Metropolitan Area (KMA) using machine learning algorithms and geospatial techniques. The investigation commenced with acquisition of the multidate Landsat imagery in specifically multidate Landsat5 TM for 1990, 2000, and 2010, and Landsat8 OLI for 2020 were utilized for the study. Following this, four Machine learning (ML) algorithms such as SVM, RF, DT and KNN are used for the LULC classification from 1990 to 2020. On the other hand, multidate spectral indices including NDVI, MNDWI, NDBSI and LST are generated using the Landsat imagery from 1990 to 2020. The integration of these four spectral indices has led to development multidate Remote Sensing-based Ecological Health Index (RSEHI) Analytic Hierarchy Process (AHP), which shows the spatiotemporal variability of the quality of ecological health in the KMA. The future simulation of LULC is conducted through Cellular Automata-Artificial Neural Network, while RESHI future prediction utilizes multivariate fractional regression techniques. The correlation matrix illustrates the relationship between LULC dynamics and RSEHI in KMA. The finding demonstrated a steady decoration in ecological health as a result of rapid urban expansion where built-up area experienced a significant increase and vegetation covers rapidly declining over time. It endorses the immediate need for sustainable ecological management strategies to address rapid ecological degradation and enhance resilience in rapidly urbanising landscapes. Machine learning and geospatial models demonstrate the rapid urbanization and large-scale spatiotemporal dynamics of land use and land cover in the Kolkata Metropolitan Area. The Remote Sensing-based Ecological Health Index (RSEHI) shows robustness in quantifying the quality of ecological health of the KMA. Urbanization and land use and land cover dynamics are strongly and inversely correlated with ecological health. The low ecological health zone is expected to increase to 1032.27 km² reflecting a significant deterioration in the ecological sustainability of the KMA till 2030.
In basins where bimodal and unimodal rainfall regimes coexist, the responses of rainfall, temperature, and drought extremes to future warming have not been adequately assessed. This study assesses future rainfall, temperature, and drought extremes of Wami basin in Tanzania using Coupled Model Intercomparison Project Phase 6 (CMIP6) multi-model ensembles. The CMIP6 outputs are statistically downscaled with the Long Ashton Research Station Weather Generator under four Shared Socioeconomic Pathways (SSP126, SSP245, SSP370, SSP585) for near (2041–2060), mid (2061–2080), and far-future (2081–2100). The selected extreme metrics from the Expert Team on Climate Change Detection and Indices (ETCCDI) and multi-timescale Standardized Precipitation-Evapotranspiration Index (SPEI-3/6/12) were analyzed separately for bimodal and unimodal rainfall regimes. The presence or absence of trends was investigated for the 1985–2014 baseline and future periods using the Mann-Kendall test and Sen’s slope estimator. The results indicate that by mid-future under high emissions, bimodal March-May rainfall increases by up to 31 The graphical abstract synthesizes the workflow, key findings, and implications of assessing future hydroclimatic change in the Wami Basin, Tanzania, a basin characterized by distinct unimodal and bimodal rainfall regimes. The left panel presents the data foundation, combining observed and ERA5 climate records with CMIP6 projections to represent both historical variability and future climate forcing. Bias-corrected statistical downscaling using the LARS-WG weather generator translates large-scale climate model outputs into basin-scale climate information suitable for impact assessment. Multi-model ensemble (MME) is constructed based on model performance. Climate extremes are evaluated through an integrated framework combining ETCCDI indices, Standardized Precipitation-Evapotranspiration Index (SPEI) drought metrics, and non-parametric trend detection using Mann-Kendall and Sen’s slope approaches across four Shared Socioeconomic Pathways (SSP126, SSP245, SSP370, and SSP585). The central panel summarizes the principal scientific findings, showing consistent warming across rainfall regimes alongside intensification of extreme rainfall events and a late-century drying tendency indicated by declining SPEI-12 conditions, particularly in unimodal areas. Spatial patterns and time-series responses highlight regime-dependent hydroclimatic behavior rather than uniform basin responses. The right panel translates these climate signals into system-level consequences, including increased flash-flood occurrence, higher evaporative losses, enhanced soil erosion, and elevated drought risk. The graphical abstract links these findings to key adaptation priorities, including flood-risk management, soil and water conservation, and heat-resilient water planning. Projected rainfall increases in both bimodal and unimodal regimes of the Wami basin. Tmax and Tmin increase markedly, reaching up to + 4.6 °C by late century under SSP585. Mid-century unimodal very-wet-day rainfall increases by up to 26 mm yr⁻¹. Diurnal temperature range shows the most consistent future trend signal. Long-timescale drought risk strengthens in the unimodal regime under SSP585.
Understanding the climatic heterogeneity of Nigeria is essential for evaluating ecosystem vulnerability, climate‑risk exposure, and hydrological stability. However, existing classifications remain limited by sparse meteorological observations and predominantly single‑variable approaches. This study aims to provide a comprehensive, data‑driven eco‑climatic regionalisation of Nigeria, explicitly addressing the lack of multivariate and validated climatic frameworks that capture compound hydro‑ecological interactions. The strong north–south hydrothermal gradients of Nigeria, together with increasing climate variability, make it a critical location for developing a refined climatic zoning framework. Conventional classifications, including threshold‑based systems, often mask intra‑regional variability and fail to resolve ecological stress patterns. This study integrates four decades (1981–2024) of MODIS vegetation stress indicators, including Moisture Stress Index (MSI), Temperature Condition Index (TCI), and Vegetation Health Index (VHI), with ERA5‑Land precipitation, temperature, soil moisture, solar radiation, and runoff datasets. Principal Component Analysis (PCA) was applied to extract dominant climatic–ecological gradients, followed by K‑means clustering to delineate homogeneous eco‑climatic regions, while quantitative validation against the Köppen–Geiger classification (Cohen’s Kappa and Adjusted Rand Index) was performed to assess robustness and added value. Five coherent climatic regimes emerged along a distinct north–south gradient. Moisture availability was identified as the primary driver of climatic differentiation, with the semi‑arid northern zones exhibiting the highest moisture stress (MSI = 0.96) and the lowest vegetation health (VHI = 39.5). Humid southern regions displayed the lowest stress levels (MSI = 0.40–0.53) and higher VHI (> 53), while the Middle Belt formed a transitional ecotone with balanced hydro‑thermal conditions. Statistical validation (ANOVA, p < 0.001) and clustering diagnostics confirm that these regions are both distinct and internally consistent, while approximately 68 Based on the graphical abstract, this study presents an integrated multivariate framework for delineating ecological zones across Nigeria using four (4) decades (1981–2024) of MODIS and ERA5-Land datasets. It begins with a map of the study area, highlighting the pronounced latitudinal hydroclimatic gradients that structure the environmental conditions in Nigeria. The workflow then introduces the two major input datasets: MODIS land surface products and ERA5 Land reanalysis variables, which were systematically harmonised by spatial resampling onto a common 0.1° grid to ensure analytical comparability. Environmental stress indices derived from MODIS (Moisture Stress Index, Temperature Condition Index, and Vegetation Health Index), together with multivariate hydroclimatic variables from ERA5-Land (precipitation, temperature, soil moisture, solar radiation, and runoff), were processed to characterise long-term climatic and eco-physiological dynamics. Principal Component Analysis (PCA) was employed to extract dominant climatic–ecological gradients, enabling substantial dimensionality reduction and revealing strong moisture-driven north–south contrasts. Subsequently, hierarchical and K-means clustering algorithms were applied to partition Nigeria into five internally coherent eco-climatic zones, ranging from arid and semi-arid domains in the north to humid tropical systems in the south, with the Middle Belt forming a distinct transitional ecotone. In general, the study presents a robust, data-driven ecological classification framework that integrates compound vegetation–temperature–moisture interactions and provides a scientific foundation for climate-risk assessment, sustainable agricultural planning, biodiversity monitoring, and environmental management across Nigeria. Developed multivariate eco-climatic zoning using 1981–2024 MODIS and ERA5-Land datasets. Validated classification vs. Köppen–Geiger (κ = 0.62; 68
Rapid urbanization has emerged as a major global environmental challenge and a key driver of air quality deterioration in metropolitan regions, particularly in rapidly developing cities. Among atmospheric pollutants, PM2.5 is of particular concern due to its severe health impacts and ability to penetrate deep into the respiratory system. This study investigates the influence of urban expansion intensity on PM2.5 concentrations in Delhi, India, by integrating urban growth indicators, demographic factors, land-use characteristics, and meteorological parameters. Multi-temporal Landsat imagery, PM2.5 observations, demographic datasets, and climatic variables were utilized to characterize urban expansion and its effects on air quality. Urban growth typologies, including infill, edge-expansion, and outlying development, were quantified using the Landscape Expansion Index (LEI), while Geographically Weighted Regression (GWR), correlation analysis, and mutual information metrics were employed to assess spatial and seasonal relationships. The urban expansion classification achieved an overall accuracy of 91.2 Based on the analytical framework, this study was conducted to examine the influence of urban expansion intensity on PM2.5 concentrations and to understand the interplay between urban growth dynamics, environmental factors, and air quality in Delhi. The work captures complex relationships among demographic, land-use, and meteorological variables within a rapidly transforming metropolitan landscape. Air quality and urban growth indicators were assessed considering both spatial and statistical controls to uncover underlying mechanisms driving pollution patterns. Techniques including correlation analysis and Geographically Weighted Regression were employed to ensure robust inference and spatial understanding. The results indicate a moderate positive association between urban expansion and PM2.5, with stronger effects in compact growth areas. Population density emerged as a dominant contributing factor, while green spaces and water bodies showed mitigating influences. Seasonal variability analysis further highlights the role of urban form in shaping pollution dynamics, offering valuable insights for sustainable urban planning and environmental management strategies.
Agriculture is one of the largest water-consuming activities worldwide and the associated groundwater abstraction is known to be a triggering factor for land subsidence. In particular, protected agriculture (PAg), although it uses more efficient irrigation systems than open agriculture, promotes more intensive overall water and soil use. However, data on groundwater abstracted volumes and piezometric data for this type of cover are not always available or reliable, offering limited opportunities to correlate them with the actual subsidence phenomena. In this study, PAg mapping was used as a proxy for groundwater abstraction in areas affected by subsidence in Zamora Valley, Mexico, an area known for berry production as PAg. A land use/land cover (LULC) and Persistent Scatterer Interferometry (PSI) analysis with Sentinel-2 and − 1, respectively, was done to obtain the spatio-temporal distribution of PAg (2019–2025) and the rates and patterns of subsidence (2018–2025). Both variables and available data on groundwater abstracted volumes and well localization were analyzed with univariate and bivariate Moran’s I, and LISA cluster maps. The LULC results (overall accuracy from 97.36 This graphical abstract shows a methodological approach for utilizing protected agriculture mapping (PAg) as a proxy for groundwater abstraction in subsidence affected areas. Particularly, the approach was implemented in Zamora Valley, Mexico, known for the production of berries under PAg, and an active subsidence from at least 14 years and recent sinking rates of 7 cm/yr. The methodology included the use of multispectral and SAR imagery to determine the spatial distribution and temporal variability of PAg, and the subsidence rates and patterns. This data, alongside available official information on PAg distribution and groundwater consumption, was analyzed with univariate and bivariate Moran’s I, and LISA cluster maps. Results reveal significant spatial variability of PAg and a poor spatial correlation with official groundwater consumption estimations. Additionally, areas with high permanent PAg were associated with greater subsidence, highlighting these areas as critical zones of sustained groundwater abstraction, and the PAg mapping as a promising proxy for water use in areas where this information is lacking. Persistent protected agriculture was associated with greater subsidence. A positive relation was detected between subsidence and wells’ density. Concessioned water volumes were uncorrelated to subsidence. The extent of protected agriculture was used as a proxy for groundwater extraction. The approach can be used in other regions with limited water-use information.
Rainfall time series are essential for hydrological and climate studies; however, data scarcity remains a persistent challenge that compromises the reliability of analyses and modeling. This study compares the performance of regression based models and machine learning (ML) methods for gap filling in monthly rainfall data from Northern Minas Gerais, Brazil, a region characterized by high climatic variability and limited monitoring infrastructure. Ten missing data levels, ranging from 5 The graphical abstract presents the workflow used to evaluate the performance of gap-filling techniques in monthly rainfall time series from Northern Minas Gerais, Brazil. The process begins with historical rainfall data, in which artificial gaps ranging from 5
Cultivation-based land suitability has gained much attention to achieve sustainability in agroforestry in the face of a changing climate, increasing rate of population growth, and shrinking arable lands with continued urbanization. The olive tree or shrub is an evergreen tree and has been of immense importance as the principal source of edible oil, providing high nutritional value, providing health benefits, and an important source of income entirely in several regions. However, little attention was paid to sustainable olive farming and the identification of suitable plantation sites. Due to the economic, social and environmental importance of olive production in Pakistan, this paper aims to consider these points and identify suitable areas for Olive plantations across Pakistan using MCDA and AHP methods in ArcGIS Software. Utilizing GIS-based multi-criteria decision analysis (MCDA) techniques, with ancillary key factors such as elevation, precipitation, temperature, soil type, slope, land use, aspect, soil pH, soil drainage, and salinity, we determined the spatial site suitable for olive plantation, and validated accuracy by field observations. In addition, we used the analytic hierarchy process (AHP) to assign proper weights for the different criteria. Results show that 37 This visual summary serves as a pivotal entry point into the research, offering a concise overview of the study’s core findings and methodologies. The graphical abstract above shows the workflow of identifying areas suitable for growing olive crops in Pakistan based on GIS MCDA using the AHP model. In the graphic, there are five main elements including: Data – the representation of the study area as well as important environmental and soil layers (elevation, slope, rainfall, temperature, land use, soil type, pH, drainage, salinity, and aspect); Analyses – showing how to proceed from resampling and normalization to overlay and suitability mapping; Model – the weightage scheme of the AHP model used to determine the weightage of the factors; Result – the result of land suitability in which 37
This study investigates the impacts of climate change on Brazil’s offshore wind energy potential across the Northeast, Southeast, and South regions using projections from 27 global climate models participating in the Coupled Model Intercomparison Project Phase 6 (CMIP6) under the Shared Socioeconomic Pathways SSP1-2.6, SSP2-4.5, and SSP5-8.5 scenarios. Wind speeds were extrapolated to 100 m and used to estimate wind power density (WPD) in offshore areas currently under regulatory assessment for wind farm development. The results reveal distinct regional responses to global warming. The Northeast maintains the highest mean WPD but shows a tendency toward reduction under the high-emission scenario (SSP5-8.5), associated with the projected weakening of the trade winds. In contrast, the Southeast exhibits a consistent increase in wind potential, particularly under SSP5-8.5, suggesting a potential intensification of the resource under stronger warming conditions. In the South, projections indicate moderate increases accompanied by considerable interannual variability. Overall, mean annual changes remain below 15 This graphical abstract presents an assessment of how climate change may affect offshore wind potential in Brazil (Northeast, Southeast, and South) using projections from 27 CMIP6 global climate models under the SSP1-2.6, SSP2-4.5, and SSP5-8.5 scenarios. The results indicate that the Northeast maintains the highest wind power density values but tends to show reductions under high-emission conditions, while the Southeast exhibits a consistent increase in wind resources and the South shows moderate increases with high interannual variability. Overall, mean annual changes remain below 15
This study investigates the impact of a mining operation on snow cover albedo in northwestern Russia. The primary objective was a spatial assessment of the pollution in the vicinity of the open-pit mine by analyzing snow albedo variations. Winter Landsat satellite images were analyzed for nine years spanning the mining activity: 1987, 1993, 1997, 2000, 2014, 2015, 2017, 2020, and 2022. Statistical analysis confirmed significant differences in albedo between snow-covered land and snow-covered ice surfaces for all studied years. Elevation, Topographic Position Index (TPI), Relative Slope Position (RSP), temperature, precipitation, wind regimes, and distances from settlements and active open-pit mines were examined as predictors for snow albedo modeling. A clear spatial dependence of snow albedo increase with distance from the pollution source was established, with the most pronounced effect observed in the early periods of 1987 and 1993. The Random Forest, Support Vector Machine (SVM), and Bagged MARS algorithms were employed to develop predictive models of spatially continuous snow albedo. The models identified distance from the mine as the most important predictor. The best performance was achieved by the Random Forest models for 2015 (RMSE = 0.04, R² = 0.76), 2020 (RMSE = 0.05, R² = 0.47), and 2022 (RMSE = 0.05, R² = 0.41). Based on the resulting models, spatial predictions of albedo were generated. The results indicate that the detectable radius of the mine’s impact on snow albedo exceeds 20 km. Spatial patterns are also controlled by local topography and prevailing winds, resulting in anisotropic pollution plumes. The spatial patterns revealed in this study and the modelled potential albedo allow for a spatially resolved evaluation of the aerogenic footprint of open-pit mining operations. The graphical abstract illustrates the concept, methodology, and key results of the study aimed at assessing the impact of a mining operation on snow cover albedo. The active open pit, drilling and blasting, material loading and unloading, as well as ore processing act as sources of dust pollution, leading to the formation of a spatial plume of aerosol particles. Snow cover albedo is used as an indicator of dust contamination. Visible albedo was derived from Landsat satellite imagery. A large part of the study area is covered by forest, which complicates the analysis of pollution plumes. To enable analysis across the entire region, random snow sampling points were generated. Based on the underlying surface type, the sampling points were grouped into snow-covered land and snow-covered ice surfaces. To assess albedo variability, environmental covariates from multiple data sources were integrated. The main groups of predictors include topographic variables (elevation, TPI, RSP), climatic parameters (temperature, precipitation, and wind regimes), distance to settlements, and distance from open pits. The machine-learning block combines several algorithms (Random Forest, Support Vector Machine, and bagged MARS). The modeling results demonstrate spatially continuous snow albedo fields, enabling assessment of the scale of impact. The detectable impact of the mining operation extends beyond 20 km. The anisotropic shape of the pollution plumes highlights the influence of local topography and prevailing winds on the spatial distribution of contamination. The impact of a mining operation on visible snow albedo in northwestern Russia was assessed using archival Landsat data. A robust spatial relationship was identified, showing increasing snow albedo with distance from the open pit. Statistically significant differences in albedo between snow-covered land and snow on ice surfaces were confirmed for all studied periods (1987,1993, 1997, 2000, 2014, 2015, 2017, 2020, and 2022). Distance from the open pit was identified as the most influential predictor of snow albedo across all predictive models. Spatial pollution patterns were shown to be anisotropic and controlled by local topography, precipitation, and wind regimes.
Urban climatology interprets cities as warmer than their surrounding areas, but in seasonally dry environments this pattern can reverse, forming oasis-like conditions. In the Brazilian semiarid, where the Caatinga responds strongly to precipitation seasonality, these patterns remain poorly documented. This study examined factors associated with the intensity of the surface thermal contrast (SHI) in 12 cities, integrating daytime Landsat Collection 2 temperatures (1996–2024), NASA POWER precipitation, MapBiomas land cover, and Embrapa soil data in Google Earth Engine, and evaluated SHI seasonality, associations with land use and land cover (LULC) classes, and variation across soil texture classes in surrounding Savanna formations. Results showed strong heterogeneity among cities. Spring presented the clearest reversal of the thermal contrast, and summer precipitation showed the most robust association with SHI in the leave-one-city-out analysis. Savanna Formation showed the lowest SHI values, while pastures and land-use mosaics showed higher values. Medium-texture soils showed the most negative SHI values in Savanna areas, but this pattern was not statistically significant. Building shadow cover (mean 8.0 This visual summary presents a synthesis of the fundamental data, analytical structure, and main results of the study. Data: The study integrates 29 years of daytime land surface temperature images from Landsat Collection 2 (1996–2024) processed in Google Earth Engine, NASA POWER precipitation records, MapBiomas land cover classifications, and Embrapa soil data for 12 Brazilian semiarid cities. Analyses: Three analytical dimensions were combined: the association between the surface heat island intensity (SHI) and hydroclimatological seasonality; the relationship between dominant land use and land cover (LULC) classes and urban-surroundings surface thermal contrasts; and the variation of SHI across soil types and textures in surrounding Savanna formations. Model: A comparative observational design at local scale was adopted, quantifying the urban-surroundings SHI index, defined as the difference between mean urban and surrounding surface temperatures, at annual and seasonal scales. Results: Urban oasis-like surface patterns were observed but are neither universal nor permanent. Spring showed the clearest median reversal; summer precipitation showed the most robust seasonal association with SHI; Savanna Formation showed lower SHI values; and medium-texture soils were associated with more negative SHI values in surrounding areas, although the textural pattern was not statistically significant. Conclusion: Urban oases in the Brazilian semiarid are best interpreted as conditional daytime surface patterns, associated with precipitation seasonality, land cover mosaic, and edaphic context, highlighting the need for integrated climate adaptation approaches in cities of arid regions. Spring presented the clearest median reversal of the urban-surroundings thermal contrast. Summer precipitation showed the most robust association with summer SHI. Savanna Formation was associated with lower SHI values than pastures and land-use mosaics. Soil texture showed exploratory, non-significant differences in surrounding Savanna areas. Urban oases emerged as conditional surface patterns, not fixed city attributes.
The city of Porto Alegre, located in the extreme south of Brazil, is one of the country’s most important urban centers. Between late April and early May 2024, it experienced the most severe flood in its history, resulting in significant infrastructural, financial, and human losses. Owing to its geographic location, several forcing factors have been identified as key modulators of the region’s climate. These include internal factors—related to oceanic and atmospheric changes—and external factors—associated with solar activity and variability in the Earth’s magnetic field. This study investigates the influence of both internal and external natural forcings on the variability of annual rainfall in Porto Alegre from 1916 to 2024. Using lag cross-correlations, spectral coherence analysis based on wavelet transforms, and partial lag cross-correlation, the research examines how climate drivers such as the El Niño–Southern Oscillation (ENSO), the Pacific Decadal Oscillation (PDO), and solar activity—including the 11-year solar cycle and the 22-year solar magnetic cycle—modulate precipitation. Results indicate that ENSO exerts the most consistent and linear antiphase influence on rainfall in the 2–7-year periodicity range. The PDO also shows persistent coherence with rainfall, particularly in multidecadal bands, often exhibiting in-phase relationships. The influence of solar cycles is more intermittent and predominantly occurs through complex mechanisms, although certain epochs demonstrate significant coherence, especially for the 22-year cycle. Additional analyses suggest that the PDO may mediate interactions between ENSO and solar activity. Overall, the findings highlight the complex interplay between terrestrial and extraterrestrial climate drivers and underscore the importance of considering both in regional hydrometeorological studies, particularly in areas affected by the South Atlantic Magnetic Anomaly. The graphical abstract represents the research entitled “Influence of Biannual to Multidecadal Time-Scale Forcings on Rainfall Recorded in Porto Alegre, Brazil.” The framework of the study is the heavy flooding that has occurred in Porto Alegre in the last one hundred years, which has caused great losses. It portrays external forcings, such as solar wind, interacting with the geomagnetic field in the region of the South Atlantic Magnetic Anomaly (SAMA). The methodology consisted of gathering data on internal forcings, represented by the Southern Oscillation Index (SOI) and Pacific Decadal Oscillation (PDO), and external forcings represented by the 11- and 22-year solar cycles, together with the annual rainfall totals of Porto Alegre. The analytical tools used were lag cross-correlation, Wavelet-based Spectral Coherence (WTC), and partial lag cross-correlation to analyze the investigated time series between the years 1916 and 2024. From the obtained results, the graphical representation leads to the conclusion that precipitation is modulated by a complex interaction between terrestrial forcing (ENSO, PDO) and extraterrestrial forcing (solar cycles). Influence of natural climate forcings on the variability of annual rainfall in Porto Alegre. Potential climatic impact of solar forcing near the center of the South Atlantic Magnetic Anomaly (SAMA). Results reinforce the significance of El Niño–Southern Oscillation (ENSO) and Pacific Decadal Oscillation (PDO) as key modulators of interannual rainfall variability in southern Brazil.
The vulnerability of forest plantations to climate change threatens global forest production, particularly as current climate commitments are unlikely to limit global warming to 1.5 °C. Developing risk indicators is essential to quantify vulnerabilities and anticipate potential losses in Eucalyptus plantations. This study developed and applied twenty-one Eucalyptus-specific bioclimatic, pests and diseases risk indicators across four global warming levels (GWLs, 1.5 °C to 4 °C). The analysis was based on high-resolution simulations (1995–2014) and projections (2015–2100) from 23 General Circulation Models from the NASA Earth Exchange Global Daily Downscaled Projections (NEX-GDDP-CMIP6), under Shared Socioeconomic Pathways SSP2-4.5 and SSP5-8.5. Daily near-surface air temperature, accumulated precipitation, relative humidity, and global solar radiation were used to calculate all indicators for a control period (1995–2014) and for time slices corresponding to each warming level. Indicators were subsequently classified into five risk levels ranging from very low to very high. Climate-driven changes in temperature, precipitation, humidity, and radiation produced broadly similar risk types across regions, differing in magnitude and spatial extent. Globally, the highest risks were associated with productivity loss (77.7–80.2 The graphical abstract presents the global risks and vulnerabilities of eucalyptus plantations under four global warming levels (GWLs). The study first identified the years at which each GWL was reached using CMIP6 General Circulation Models (GCMs). These years were subsequently linked to their high-resolution counterparts from the NEX-GDDP-CMIP6 dataset. A total of 21 indicators were developed and analyzed to quantify climate and phytosanitary risks to eucalyptus plantations at the global scale across all GWLs (1.5 °C to 4 °C). In addition, key adaptation measures were identified to mitigate these risks based on high-resolution climate projections. The results indicate that the greatest risks were associated with productivity loss (77.7–80.2
