Climate risks are systematically under-observed in the world’s most vulnerable regions, compounding exposure by delaying preparedness and adaptation. Using 2.1 million news articles across 184 countries and 40 languages, combined with AI-assisted geolocation and human-in-the-loop validation, we expose global under-reporting of extreme heat events. We find that only a tiny fraction of identified extreme heat events enters widely used global disaster databases, while 63.38% remain unreported even across the most comprehensive global and local news coverage. Under-reporting is concentrated in places like equatorial Africa, the Amazon basin, small island states, such as the Caribbean, and polar adjacent regions, where limited early warning capacity, high vulnerability, and linguistic distance from dominant languages intersect. Under the high-emissions scenario SSP5-8.5, by 2050, the largest future heat burdens are concentrated in highly vulnerable countries that coincide with the most severe under-reporting of extreme heat. These countries are projected to incur economic losses 60.09% higher and mortality-related losses 46.40% higher than better-reported countries. Yet pervasive under-reporting suggests that even these severe loss projections remain conservative for regions with the highest vulnerability, reinforcing the urgency of loss-and-damage action.

Heatwaves pose an increasing threat to public health under climate change. Despite evidence that health systems in high-latitude countries are insufficiently prepared for extreme heat, few studies have investigated the state-of-the-art deep learning (DL) models to forecast heat-related morbidity at seasonal lead times. This study develops and evaluates a multivariate, multi-step impact-based forecasting framework across Sweden for predicting heat-related morbidity using Neural Basis Expansion Analysis for Time Series (N-BEATS) models. N-BEATS models are developed and tested under recursive and multi-input–multi-output (MIMO) multi-step forecast strategies and compared with statistical baselines (ARIMA, naïve seasonal) and classical DL model (Long Short-Term Memory (LSTM)). Forecasts are generated using morbidity counts alone and in combination with exogenous covariates (Heat Wave Index and the number of individuals with respiratory diseases) while local and global modeling approaches are examined.Results show that N-BEATS with both covariate and local modelling strategy significantly outperforms all baseline models with the lowest MAE, RMSE, and MASE values. N-BEATS shows greater data efficiency with iteratively refined residuals through fully connected backcast and forecast stacked blocks compared to LSTM, particularly when there is an extreme morbidity peak. Individually trained local N-BEATS models are more effective than the cross-learning global N-BEATS, even with similar seasonal peaks and lower data quantity. Regional differences in climate, hydrology, and demographics could hinder the effectiveness of global models and underscore the importance of localized adaptation plans and measurements. Models may also underperform during unprecedented periods, such as during the COVID-19 pandemic in 2021. The underperformance may have resulted from disruptions in healthcare during COVID, behavioral change from seeking healthcare, and selected covariates didn’t capture healthcare system capacity. Future study could be improved by testing model performance to incorporate a covariate that reflects healthcare system capacity, such as service load to enhance model’s robustness to similar system level shock.The study offers a concrete step toward operational impact-based early warning systems by enabling national agencies to anticipate heatwave burdens when a seasonal heatwave alert is issued. By coupling hazard forecasting with health impact prediction, this work supports the development of impact-based early warning systems tailored to the growing risks of extreme heatwaves. Integrating morbidity forecasts into heat-health action plans can support public health agencies in proactive resource allocation, risk communication, and preparedness planning.

More and more people are moving to cities, which means cities play a big role in both causing and solving climate change. Cities produce about 70% of global greenhouse gases (GHGs), so finding ways to reduce these is key to reaching climate change goals. One way to reduce GHGs is through nature-based solutions—using nature, like trees, parks, and green roofs, to make cities greener and healthier. Nature-based solutions can lower GHGs by storing carbon, cooling the air, and reducing energy use. They also make cities better places to live by cleaning the air, reducing floods, and supporting biodiversity. In this study, we looked at cities in Europe to understand how much nature-based solutions can reduce GHGs. We found that nature-based solutions could cut city GHGs by an average of 17%, and up to 25% in some cities. Working with nature can help cities reach their climate goals faster and more sustainably.

The primary objective of this study was to investigate, for the first time on a global scale, the effects of spatial and temporal variability in climate and land use on the occurrence of floods in the twenty-first century using different shared socioeconomic pathway-representative concentration pathway (SSP-RCP) scenarios (i.e., SSP1-RCP2.6 and SSP5-RCP8.5). Uncertainties in climate and land-use change projections were considered using an average of 13 global climate models (GCMs) and the Land-Use Harmonization 2 (LUH2) dataset. Eight dynamic precipitation variables, seven land-use fractions, and seven topographic factors were used to generate flood susceptibility maps based on different current and future scenarios (2041–2060 and 2061–2080). Predictive models were developed using three machine learning algorithms (regularized logistic regression, boosted classification tree, and random forest [RF]). The predictive ability of these models was assessed in terms of sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and area under the receiver operating characteristic curve (AUC). The results indicated that RF outperformed the other models and had the highest predictive ability during the testing phase (sensitivity = 0.91, specificity = 0.86, PPV = 0.88, NPV = 0.90, and AUC = 0.96). In all current and future scenarios, the results revealed the impacts of land use and climate change in expanding the area characterized by high/very high flood susceptibility, particularly in Oceania (New Zealand, Fiji, Guam, and the Solomon Islands), Europe, and several Asian and African countries. Relative importance analysis based on the results of the best-performing model (RF) revealed the significant role of managed pasture, urban land, and precipitation-related variables in mapping flood-prone areas, while topographic variables played a comparatively minor role. These results provide a more accurate basis for assessing and mitigating the impact of floods, enhancing resilience, and ensuring public safety by considering future changes in land use and climate conditions.

Floods are renowned as the most destructive natural phenomena, and their frequency and intensity increase due to climate change. Accurate and timely flood mapping is critical for effective risk mitigation. However, traditional approaches relying on optical remote sensing imagery and synthetic aperture radar (SAR) classification face significant limitations due to cloud cover and misclassification-induced low accuracy. To address these challenges, this study developed a novel hybrid framework of metaheuristic optimization (MO) and deep learning (DL)-based semantic segmentation for more precise flood mapping. Three MO algorithms including artificial bee colony (ABC), genetic algorithm (GA), and swarm-based simulated annealing (SwarmSA) were used to identify the most informative combination set of polarimetric SAR (PolSAR), including VV and VH, PolSAR decomposed features, and textural descriptors from the Gray-Level Co-occurrence Matrix (GLCM). Three convolutional neural network (CNN)-based DL models (e.g., VGG16-U-Net, DRN, and CPNet) were trained to extract flood inundated areas from the Sentinel-1 SAR imagery. The proposed methodology was applied to the April 2019 flood event in Khuzestan province, Iran. The results showed that the CPNet model coupled with the SwarmSA achieved the highest F1-score (flooded areas: 0.901; non-flooded area: 0.976) and IoU (flooded areas: 0.820; non-flooded area: 0.954) in mapping inundated areas. Furthermore, the selected feature set, which includes dissimilarity from VV, GLCM correlation from VV, homogeneity from VH, GLCM mean from VV, and GLCM correlation from VH effectively captured the spectral and textural characteristics of flooded areas. The results highlighted the effectiveness of integrating MO-based feature selection techniques with DL architectures to achieve high-resolution and expeditious flood extent mapping.

Although tools for climate change adaptation have proliferated, there is relatively little evidence about who they are intended to serve and how well they are suited to respond to decision contexts. We synthesized an inventory of 122 tools and found that around 73% of these are relevant to adaptation in the Mediterranean. We then examined who they are designed for, what they support, and how well they are suited to respond to decision contexts. We argue that access is not the main challenge, but alignment is. The results show that tools operate in isolation, often bound to administrative rather than physical boundaries, and provide limited guidance for choosing appropriate applications. Multilingual support and pathways to integrate local data are uneven. We outline actionable directions to improve this developing ecosystem, including linking tools to each other and to planning processes, making assumptions explicit, and involving anticipated users in the design process. These steps can turn a large and growing supply of tools into a more coherent, context-aware, and usable resource for adaptation and planning across the Mediterranean. Climate change adaptation tools in Mediterranean regions often fail the people and places that need them most, because they remain expert-oriented, administratively bounded, and poorly matched to regional realities, according to systematic evaluations.

With the intensification of climate change and anthropogenic activities, water scarcity and drought have become critical challenges around the world, threatening various ecosystems, particularly forests. Forests are social-ecological systems that provide numerous services to humans, who, in return, alter them. While it is impossible to prevent droughts, understanding the attributes of forests, particularly their resilience, may facilitate the mitigation of drought-related adverse consequences. Resilience is a multifaceted concept that has been interpreted through various lenses in the literature, with engineering resilience emphasizing system recovery, ecological resilience investigating the adaptive capacity of forests, and social-ecological resilience highlighting the interconnectedness of human and natural systems in resilience assessment.Building on these conceptual foundations, seven principles of resilience, maintaining diversity and redundancy (P1), managing connectivity (P2), managing slow variables and feedback (P3), fostering complex adaptive system thinking (P4), encouraging learning and experimentation (P5), broadening participation (P6), and promoting polycentric governance (P7) offer a comprehensive approach to building, evaluating, and enhancing resilience. This review aims to investigate the extent to which resilience principles have been integrated into the discourse of forest resilience to drought in the literature.Searching the Web of Science database for studies on forest resilience from 1998 to 2024 resulted in 47 papers. Among the reviewed studies, 51% investigated resilience through the lens of ecological resilience, 30% utilized the social-ecological concept, and 19% employed engineering resilience. P4 is frequently examined using tree ring data and drought severity indices (e.g., SPEI). Species richness and composition have often been considered to evaluate P1. A close examination of the methodologies of the reviewed studies revealed that 34% are evidence-based or conceptual studies aimed at understanding the mechanisms contributing to resilience, and 21% are experimental and field studies, which often involve the use of collected field data, such as tree ring width, vegetation growth rate, to explore the response of forest systems to natural or experimentally induced drought events.The limited use of modeling, specifically landscape or ecosystem services models, in studying forest resilience to drought is evident, with only three studies conducted on this topic. Furthermore, the case studies are nearly evenly distributed across Africa, Europe, North America, and Asia, with 7, 10, 10, and 8 studies, respectively. Four studies investigated the resilience of forests in South America, and another four focused on a global scale. A closer exploration of the reviewed studies revealed that no studies have attempted to consider all seven resilience principles jointly, highlighting a significant research gap in this area and emphasizing the need for more studies to tackle the intricate relationships between ecosystems and human communities and societies.

Record-breaking temperatures and frequent heatwaves have been experienced worldwide in recent years. Heatwaves pose an escalating threat to public health and heat-related impact forecasting is critical to implementing suitable mitigation strategies. Deep learning (DL) models, notably Long Short-Term Memory (LSTM), have been widely applied for heat-related impact forecasting. However, the emergence of state-of-the-art forecasting DL architectures such as Neural Basis Expansion Analysis for Interpretable Time Series Forecasting (NBEATS) provides a novel solution for long-term heat-related impact forecasting. This study develops, evaluates, and compares multiple time series forecasting models-including advanced DL architectures (N-BEATS, N-HiTS, LSTM), a classical statistical model (ARIMA), and a Na & iuml;ve seasonal baseline-to predict heat-related morbidity across 21 Swedish counties using data from 2008 to 2023. Both local (individually trained) and global (crosslearning across counties) modeling strategies were explored, incorporating exogenous variables (Heatwave Index and number of people with respiratory disease), and comparing recursive and Multi-Input-Multi-Output (MIMO) forecasting output strategies. Results indicate that the local N-BEATS model achieves superior predictive performance, particularly when both exogenous variables are included. MIMO generally yields a better performance by mitigating error propagation over extended forecasting horizons. Moreover, individually trained N-BEATS models outperform cross-learning global N-BEATS, underscoring the importance of localized adaptation plans. These findings highlight the potential utility of multivariate N-BEATS for more accurate heatwave impact forecasting. This study can complement and support early warning frameworks by integrating the developed impact forecast model with existing hazard models, thereby enabling more proactive public health interventions and improving community resilience to heatwaves.

This study examined global atmospheric variations in carbon dioxide (CO2) and methane (CH4), which are two major greenhouse gases (GHGs). The main objectives were as follows: (1) identify the top 50 cities with the highest long-term (2003-2020) mean annual CO2 and CH4 mixing ratios (kg/kg) at 1000 hPA, (2) analyze 18year trends in mixing ratios for cities worldwide, (3) conduct a spatiotemporal analysis of column-averaged CO2 and CH4 concentrations, and CO2 and CH4 mixing ratios at 1000 hPa from 2003 to 2020, and (4) assess the impact of GHG emissions on near-surface air temperature (NSAT). These objectives were achieved using the Copernicus Atmosphere Monitoring Service (CAMS) global greenhouse gas reanalysis product and the ERA5 reanalysis dataset. The findings indicated that Chengdu (China) had the highest long-term mean annual CO2 mixing ratio, followed by Luoyang (China), Chongqing (China), Myitkyina (Myanmar), Louangphrabang (Laos), Lampang (Thailand), Louang Namtha (Laos), Aizawl (India), Nola (Central African Republic), and Los Angeles (USA). Katowice (Poland) exhibited the highest long-term mean annual CH4 mixing ratio, followed by Sao Paulo (Brazil), Lahore (Pakistan), Delhi (India), New Delhi (India), Moscow (Russia), Chengdu (China), Anshan (China), Andijan (Uzbekistan), and Fergana (Uzbekistan). Between 2003 and 2020, the mean annual columnaveraged CO2 concentration increased from approximately 372-379 ppm to 409-418 ppm, whereas the column-averaged CH4 concentration increased from about 1599-1861 ppb to 1711-2016 ppb. The global average CO2 and CH4 concentrations (weighted by the cosine of latitude) increased from nearly 375 ppm to 412 ppm and from 1735 ppb to 1837 ppb, respectively, during 2003-2020. The global average NSAT (weighted by the cosine of latitude) also increased from 14.36 degrees C to 14.80 degrees C between 2003 and 2020, showing an increase of 0.44 degrees C. This comprehensive analysis highlights the urgent need to address GHG emissions to mitigate their environmental and climatic effects.

Monitoring urban expansion, particularly the detection of new building constructions, is crucial for sustainable planning and management. This study assessed the effectiveness of Sentinel-2 satellite imagery for detecting new constructions in Mehriz County, Yazd Province, Iran, where population growth has accelerated conversion of agricultural land to urban areas. Sentinel-2 images and fused Sentinel-2 datasets with Unmanned Aerial Vehicle (UAV) imagery, with and without the inclusion of a Digital Surface Model (DSM), were used to detect building constructions between April 2023 and April 2024, using classification methods. The classification accuracy of five pixel-based and five object-oriented classification algorithms was investigated for classifying the dataset on the two acquisition dates. The results show that the pixel-based Maximum Likelihood (ML) classifier and the object-based K-Nearest Neighbors (KNN) classifier showed the highest accuracy in land cover classification for the April 2023 and 2024 Sentinel-2 images, with Kappa coefficients of 0.67 and 0.65, respectively. Fusion of Sentinel-2 and UAV imagery using the Gram-Schmidt method improved classification accuracy. For fused Sentinel-2 images without DSM, the Support Vector Machine (SVM) achieved the highest overall accuracy and Kappa coefficient in pixel-based classification for both April 2023 (82.62% and 0.76), and April 2024 (80.77% and 0.74). Among object-based methods, the SVM and KNN algorithms, with similar performance, yielded overall accuracies of approximately 84.0% (Kappa = 0.78) for 2023 and 82.0% (Kappa = 0.76) for 2024. Incorporating DSM data further improved classification results by 1–2% across some pixel-based and object-based classification methods. According to the change detection maps generated from the SVM (pixel-based) and KNN (object-based) classifiers, 61.9% and 72.4% of new building construction points were correctly identified from the total number of 410 ground-truth construction points between April 2023 and 2024, respectively. These findings demonstrate that fused Sentinel-2 satellite imagery offers a cost-effective and reliable approach for continuous monitoring of urban fringe development.

This study investigates the influence of climate variables, specifically temperature and relative humidity, on the equilibrium moisture content (EMC) of wood-a critical quality parameter. Using data from 100 synoptic stations across Iran (1987-2019), we analyzed trends in temperature, humidity, and EMC through the Mann-Kendall and Sen's slope methods. Future projections (2020-2049) employed CMIP6 models-CanESM5, CanESM5-CanOE, CNRM-CM6-1, CNRM-ESM2-1, and IPSL-CM6A-LR-under SSP scenarios, with model selection based on RMSE, Scatter Index, and R². Scenarios SSP1-2.6, SSP2-4.5, and SSP5-8.5 were used to project future climatic conditions and corresponding EMC values. The CanESM5-CanOE model exhibits the lowest monthly relative humidity estimation errors in Iran, with errors ranging from 10.1% to 15.0% across different climate zones. Increasing EMC is most frequent under SSP1-2.6 (20%-92% of stations) and SSP5-8.5 (34%-100%). Decreasing trends are significant under SSP2-6.5 (66%-100%) and SSP5-8.5 (45%-88%). Monthly variations: -4.74% to + 3.71%; seasonal: -2.87% to + 2.45%; annual: -1.17% to + 1.00%. Significantly decreasing EMC trends are under SSP2-6.5, increasing trends under SSP5-8.5. Over a 30-year span, EMC varied from 0.06 to 0.62% in winter, from - 1.14 to -1.23% in spring, from - 0.84 to -0.89% in summer, and from - 0.80 to -1.34% in autumn, with most changes being statistically significant. These findings suggest climate change will substantially impact on wood EMC, underscoring the importance of revising future EMC standards accordingly.

Quantifying spatial landscape patterns along urban-rural gradients is crucial for understanding urbanization's impacts on ecosystem structure and function. This study presents an integrated framework assessing land use patterns and ecological quality in the Greater Isfahan region, Iran. A novel direction-specific Ecological Quality Index (EQI) for green cover is introduced to describe spatial patterns of urban expansion and associated ecological conditions. Methodologically, we classified Landsat 9 imagery into six land cover types using Support Vector Machine (SVM) and achieving 94% accuracy. Gradient analysis was conducted along eight geographical transects, partitioned into 3 × 3 km blocks. Five key metrics (NP, AREA_MN, ENN_MN, LPI, ED) quantified fragmentation, connectivity, and dominance. Kruskal-Wallis tests confirmed significant differences along transects (p < 0.001). Cluster analysis identified seven distinct classes, ranging from dense urban areas to rural landscapes. The EQI was derived through Principal Component Analysis (PCA) using green cover related metrics, utilizing factor loadings and eigenvalues from the first two components. Results reveal a clear divergence in landscape structure. The mean patch area of impervious surfaces shows a decreasing gradient, from over 28 ha in central blocks to less than 1 ha in peripheral zones. Conversely, agricultural mean patch size increases to 27 ha in northern transects. The EQI analysis highlights spatial inequities. The northeast-southwest transect exhibits higher ecological sustainability, with high-quality areas constituting 20% of the landscape, contrasting with less favorable conditions along the west-east transect.This study demonstrates that integrating gradient-based metrics with a direction-specific EQI provides a powerful, replicable tool for diagnosing urban ecological health. The findings offer actionable insights for targeted planning, emphasizing agricultural protection and green infrastructure connectivity to mitigate fragmented urbanization in arid-region cities.

Understanding the total water flows and pollutant loads to the Baltic Sea is important for effective coastal-marine ecosystem management. Current assessments often overlook the unmonitored flows and submarine groundwater discharge (SGD). This study proposes and outlines a conceptual modelling framework for overcoming this common neglect by integrated quantification of (1) the monitored surface water flows, and the unmonitored (2) surface water flows and (3) SGD from land to the Baltic Sea. The study outlines how unmonitored runoff and SGD can be estimated by various quantification approaches based on commonly available hydro-climatic, hydrogeological, and other characteristic catchment data. It also describes how modules for the different monitored and unmonitored discharge components are linked and should be integrated in modelling to total annual, seasonal, or finer-resolved water flows to the Baltic Sea, and analogously also in other coastal regions around the world. Though quantitative modelling remains ongoing, the conceptualization opens pathways to improve assessments and management of freshwater flows and associated pollutant loads to the Baltic Sea.

This study examined global atmospheric variations in carbon dioxide (CO2) and methane (CH4), which are two major greenhouse gases (GHGs). The main objectives were as follows: (1) identify the top 50 cities with the highest CO2 and CH4 emissions, (2) analyze 17-year trends in emissions in cities worldwide, (3) conduct a spatiotemporal analysis of CO2 and CH4 emissions from 2003 to 2020, (4) quantify changes in GHG emissions during this period, and (5) assess the impact of GHG emissions on land surface air temperature (LSAT). These objectives were achieved using the global ERA5 reanalysis data from the Copernicus Climate Change Service. The findings indicated that Chengdu (China) had the highest cumulative CO2 emissions between 2003 and 2020, followed by Luoyang (China), Chongqing (China), Myitkyina (Myanmar), Louangphrabang (Laos), Lampang (Thailand), Louang Namtha (Laos), Aizawl (India), Nola (Central African Republic), and Los Angeles (USA). Katowice (Poland) exhibited the highest CH4 emissions, followed by São Paulo (Brazil), Lahore (Pakistan), Delhi (India), New Delhi (India), Moscow (Russia), Chengdu (China), Anshan (China), Andijan (Uzbekistan), and Fergana (Uzbekistan). Between 2003 and 2020, the mean annual atmospheric CO2 concentration increased from 394.470–394.477 ppm to 394.501–394.510 ppm, whereas the CH4 concentration increased from 1831.3–1833.3 ppb to 1832–1834.5 ppb. The analysis revealed significant increasing trends in CO2 and CH4 emissions globally, with certain cities exhibiting sharper increases. The LSAT also increased during the study period, with the minimum LSAT increasing by 2 °C (from − 54.49 °C to − 52.20 °C). This comprehensive analysis highlights the urgent need to address GHG emissions to mitigate their environmental and climatic effects.

With the acceleration of global urbanization, the ecosystem services (ES) and ecological balance of nature reserves have been significantly impacted. However, quantitative assessments of the multiple contributions of nature reserves to urban ecological sustainability are still lacking. This study selects Panjin, a wetland city in China (3788 km2), as the study area, utilizing the InVEST model to quantify ES (water yield, carbon storage, soil retention, and habitat quality), and employing redundancy analysis to explore the influencing factors. Ecological source areas were identified, and the Sustainable Development Goals (SDGs) score was calculated to systematically evaluate the contribution of nature reserves. The results indicate that from 1990 to 2010, the built-up area of Panjin increased by approximately 159%, leading to a reduction in carbon storage, soil retention, and habitat quality by 20%, 4%, and 14%, respectively. From 2010 to 2020, ecological restoration policies resulted in a 63% increase in ES compared to 2010. Nature reserves played a crucial role in maintaining ecological stability, providing over 40% of the ecological source areas while occupying only 24% of the city’s area and contributing more than 30% to the overall urban ecological sustainability. This study is the first to systematically assess the multiple contributions of nature reserves to urban ecological sustainability, providing ecological management recommendations for policymakers based on innovative environmental indicators and methods to support sustainable urban development.

In recent years, the wide availability of high-resolution radar satellite images has enabled the remote monitoring of wetland surface areas. Machine learning models have achieved state-of-the-art results in segmenting wetlands from satellite images. However, these models require large amounts of manually annotated satellite images, which are slow and expensive to produce. The need for annotated training data makes it difficult to adapt these models to changes such as different climates or sensors. To address this issue, we employed self-supervised training methods to develop a model, AquaCluster, which segments radar satellite images into water and land areas without manual annotations. Our final model outperformed other radar-based water detection techniques that do not require annotated data in our test dataset, having achieved a 0.08 improvement in the Intersection over Union metric. Our results demonstrate that it is possible to train machine learning models to detect vegetated water from radar images without the use of annotated data, which can make the retraining of these models to account for changes much easier.

This study investigates how the seven core resilience principles are integrated into assessments of forest system resilience to natural or human-induced disturbances across engineering, ecological, and social-ecological resilience concepts. Following PRISMA guidelines, a literature search in the Web of Science database using the keywords "resilience", "forest" and "ecosystem services" yielded 1828 studies, of which 330 met the selection criteria. The most commonly used criterion was diversity, a sub-criterion of "diversity and redundancy", appearing in 50% of studies. The results indicate that social and governance-related principles, learning and experimentation (7%), participation (11%), and polycentric governance (9%) have not been frequently addressed. Although numerous studies have employed various principles for assessing forest resilience, none have considered all seven principles jointly. This highlights a significant research gap, emphasising the need to quantify these principles in forest systems. Understanding forest-community dynamics is essential for enhancing the long-term resilience and sustainability of both systems.

Over the past decade, water conflicts have risen, and cooperation has declined. Research highlights multiple factors driving this change, with climate change acting as a threat multiplier. Human activities, like dam construction and irrigation, and climate-induced hydro-climatic shifts, including extreme precipitation and prolonged droughts, contribute to the risk of increased water conflicts. To guide interventions and reverse this trend, our focus is on enhancing the understanding of factors that facilitate successful cooperation and mitigate water conflicts effectively. In this study, we investigate cooperation and conflict events worldwide in the last 70 years, together with climatic and socioeconomic factors, such as wealth, export dependency, demographics, water use, and hydro-climate trends. The dataset on cooperation and conflict events used is based on the Transboundary Freshwater Dispute Database and Water Conflict Chronology in combination with more current cooperation events extracted from media news reports. Relationships between investigated factors and cooperation are analyzed by combining panel data analysis and qualitative text content analysis of events. The results provide a deeper understanding of the factors behind why certain events are more successful in achieving conflict mitigation than others. We found that cooperation between countries struggling with water-related challenges can reduce expected conflicts over the next five years. The economic benefits of cooperation show a positive correlation between water-related cooperation and improved wealth (measured by GDP growth), particularly in countries with high export dependency. As such, economic collaboration can be an effective tool for enhancing resilience in high-water stress areas, where collaboration in these areas can contribute to a substantial reduction in future conflicts while simultaneously improving economic prosperity. Engaging in cooperation with other countries can therefore contribute to economic growth and resilience, as well as decreasing conflict risk. Understanding successful conflict mitigation factors can provide helpful insights to global policymakers and leaders in water management to avoid future conflict based on current and projected water availability. Keywords: water conflict; collaboration; conflict mitigation; mixed methods; socioeconomic factors