Interaction networks between frugivorous birds and fruit trees are essential for sustaining bird diversity and supporting their ecological roles in urban environments. However, little is known about how these networks differ across various urban sites. This study analyzed these interaction networks in 24 Beijing urban parks to explore their structural differences. Results indicated that networks are generally simplified, mainly influenced by a few widespread generalist species. Nonetheless, network beta-diversity analysis showed significant differences among parks, with high dissimilarity (beta WN) driven by species turnover (beta ST) and changes in species composition (beta S). Modularity (Q) was a key factor in grouping parks, and features such as vegetation cover and fruit tree diversity were significantly associated with network variation. These findings highlight the variability of bird-fruit tree interaction networks within cities and suggest that maintaining complex, diverse networks in urban areas is achievable.

Understanding how temperature, an environmental factor associated with suicide and homicide, may assist in identifying preventive approaches at the population level. However, significant gaps remain in understanding this association in tropical developing regions with high homicide and suicide burdens, where underlying socio-structural factors may critically modify temperature effects. By integrating 1,243,596 records extracted from two national mortality information systems and daily weather data between 2000 and 2019, the relationship between increased ambient mean temperature and the risk of suicide and homicide mortality was investigated using a time-stratified case-crossover design. Odds ratios and 95% confidence intervals (CIs) of temperature-associated mortality were estimated using the conditional logistic regression model combined with the distributed lag nonlinear model. Pooled estimates suggested each 1 °C increase in ambient mean temperature was associated with a 2.17% (95% CI 1.72–2.63%) and 1.80% (95% CI 1.33–2.27%) increase in homicide over lag 0–5 days in Brazil and Mexico, and the estimates for suicide were 2.78% (95% CI 2.18–3. 39%) and 3.75% (95% CI 3.05–4.46%), respectively. The association appeared stronger among males, mixed-race individuals in Brazil, during weekends, and in summer–autumn. Socioeconomic factors significantly modified the association between temperature and risks of suicide and homicide mortality. Throughout the study period, 10.19% of suicides and 9.43% of homicides in Brazil, along with 12.97% of suicides and 12.56% of homicides in Mexico, were associated with temperatures exceeding state-specific medians. Our findings emphasize the significance of heat adaptation strategies throughout almost the year to mitigate self-harm and interpersonal violence. In this study, Huang et al. link rising ambient temperatures to increased suicide and homicide mortality in Brazil and Mexico, showing stronger effects in specific groups and contexts and highlighting the role of socioeconomic factors and the need for heat adaptation strategies.

Diverse tree communities can bolster urban ecosystem resilience and provide vital ecosystem services. However, existing urban tree species datasets have limited geographic coverage and contain inadequate attributes. To address those gaps, we developed the Global Urban Tree Species (GUTS) dataset by integrating data from literature, biodiversity databases, and other open sources. The new dataset encompasses 159,845 occurrence records of 10,094 tree species in 8,349 cities and 139 countries. Among them, 109,879 records were confirmed from urban areas, representing 11.18% of global tree species diversity. The dataset has been validated using multiple methods. GUTS fills critical data gaps and provides a foundation for future research and management of global urban biodiversity.

Remote sensing analysis converts remotely sensed data into actionable insights across diverse fields. However, traditional data-driven approaches frequently overlook the technical challenges encountered by end-users lacking remote sensing expertise. To address this issue, we proposed a multi-agent system framework based on a large language model, designed to facilitate remote sensing analysis for non-experts. The framework consists of three main modules: Data, which includes remote sensing datasets and supplementary information; Tools, encompassing algorithms and visualization tools; and Brain, which provides AI-driven task management and reasoning. We employed a prototype system called ExpertsRS, developed on the AutoGen framework and DeepSeek-V3, to validate the proposed framework. This prototype features three LLM-powered agents with distinct roles and functionalities, collaborating to fulfill user requests. The system was evaluated through two experiments. The first experiment demonstrated that ExpertsRS could translate ambiguous queries into structured analytical workflows and produce outputs comparable to those of human experts. The second experiment indicated that ExpertsRS outperformed both a baseline LLM and a single-agent configuration in planning efficiency and result accuracy, with only a modest increase in token usage. Our system underscores the potential of large language model-based multi-agent system to assist end-users in overcoming technical barriers in remote sensing analysis.

BACKGROUND:The spatial distribution of tree canopies influences ecological functions and residents' exposure to green spaces. Although several studies have examined green space configuration at neighborhood scales, evidence on tree canopy configuration at the municipal scale, an operational unit for urban planning, remains limited. METHODS:We conducted a nationwide ecological study of 2,136 Swiss municipalities. Tree canopy coverage (PLAND), aggregation (AI, reflecting how tightly green patches are grouped together), patch density (PD, a measure of fragmentation), and area-weighted mean shape index (SHAPE_AM, a measure of shape complexity) were derived from 1-m canopy maps within municipality-specific populated areas. Natural-cause, cardiovascular, and cancer mortality (2017-2019) were obtained from the Swiss National Cohort. Fully adjusted negative binomial regression models estimated associations between canopy metric and mortality for each IQR increase in the metrics. RESULTS:Holding configuration constant, each IQR increase in canopy coverage (∼18%) was associated with a 3.6% [B: -0.036; 95% CI: -0.078 - 0.005] reduction in cardiovascular mortality. Higher aggregation corresponded to a 4.3% [B: 0.043; 95% CI: 0.026-0.061], an 8.9% [B: 0.089; 95% CI: 0.059-0.119], and a 2.1% [B: 0.021; 95% CI: 0-0.042] higher number of natural-cause, cardiovascular, and cancer deaths respectively. Higher fragmentation was associated with a 3.3% [B: 0.033; 95% CI: 0.016-0.050], a 4.9% [B: 0.049; 95% CI: 0.020-0.078], and a 2.2% [B: 0.022; 95% CI: 0.001-0.043] increase in these causes respectively. No meaningful associations were observed between shape complexity and any mortality outcomes. Associations for aggregation and fragmentation were generally stronger in highly urbanized municipalities. CONCLUSIONS:At the municipal scale, mortality was lower where tree canopy was distributed across several moderately sized, spatially balanced patches rather than highly aggregated or highly fragmented structures. These findings suggest that urban greening strategies should optimize its spatial configuration to maximize health benefits.

OBJECTIVES:Despite rotavirus (RV) being the leading viral cause of childhood diarrhea in China, burden estimates and severity profiles associated with natural infection remain limited. We aimed to quantify RV burden and clinical progression among unvaccinated Chinese infants using vaccine clinical trial data. METHODS:We pooled efficacy and immunogenicity data from placebo arms of RV clinical trials conducted in China. A six-state burden pyramid (susceptible, infected, symptomatic, severe, hospitalized and dead) described the natural history. We used a unified Bayesian framework to estimate transition probabilities between adjacent states and derived RV-associated burden and clinical severity metrics. Sensitivity analyses restricted the dataset to trials with > 3000 observations and to infants enrolled at 6-16 weeks. RESULTS:We included 20,248 observations from 11,235 eligible infants. We estimated an annual RV infection rate of 495.8 (95% credible interval [CrI]: 314.1-782.8) per 1000 infants. Only 10.3% (4.8-21.8), 4.1% (2.0-8.5), 1.5% (0.8-2.9) and 0.09% (0.02-0.5) of infections progressed to symptomatic, severe, hospitalized and fatal cases over one year. CONCLUSION:Natural RV infection is frequent but predominantly subclinical, indicating that surveillance based on severe or hospitalized cases underestimates the true burden. Our findings support RV vaccine introduction in China's national immunization program.

[Objective]Creating bird-friendly urban green spaces is a tangible step toward fulfilling the national biodiversity conservation strategy and has gained attention in urban greening efforts.Developing standardized principles and core guidelines for these spaces can help guide construction of green spaces across different areas.Nonetheless,there is limited research on what should be incorporated into guidelines for building bird-friendly green spaces. [Methods]Focusing on the principles and content of the guidelines for building bird-friendly green spaces,this study used a literature review to summarize patterns of avian diversity in urban green spaces and their influencing factors.It then examined the resources birds need for survival and the risks they face in these environments.Based on the analysis,essential principles and a content checklist for guideline development have been proposed.The checklist was used to analyze the contents of relevant standards,technical guides,and guidelines to identify their strengths and weaknesses.Additionally,aspects requiring attention in future efforts to develop guidelines were suggested. [Results]Urbanization has altered bird diversity in cities both spatially and temporally,shaping overall patterns such as distribution along urban-rural gradients,species composition,and abundance of urban birds.These patterns are mainly driven by environmental filtering and human activities.Habitat changes and pollution—air,light,water,noise—filter urban birds,while human persecution and the introduction of new species select for those adapted to city life.Because bird diversity in urban green spaces is influenced by various factors across different scales,scale-based planning and management are crucial for creating effective bird habitats in green spaces.A truly bird-friendly urban green space must provide essential resources like food,water,shelter,and nesting sites,while reducing risks from light and noise pollution,domestic cats,and human disturbance.Guidelines should be based on core principles that provide overall direction,evaluate reasonableness,and adapt to unforeseen circumstances.Recognizing that making urban green spaces bird-friendly is a multidisciplinary conservation effort,five principles are proposed.1)Multi-scale planning and design.2)Prioritizing protection over new construction.3)Considering species-specific traits.4)balancing resource provision and risk management.5)Offering multiple ecosystem services.Building on these principles,a checklist with 28 items covering city,district,and site scales is presented to guide the development and assessment of green spaces.Comparing existing guidelines to the checklist shows that all include site-scale construction techniques,indicating a focus on detailed project guidance.Several guidelines also address district-scale measures,with some extending to city-scale considerations,reflecting a move towards systematic,multi-scale planning to enhance urban bird diversity.All guidelines mention the design and post-implementation phases,but few detail the implementation process itself.Explicit impact mitigation during construction remains underdeveloped.At the city scale,existing guidelines focus on urban planning,goal setting,zoning,and artificial facilities but lack comprehensive surveys and community participation.High technical and resource demands limit city-wide bird surveys,resulting in little data on urban bird diversity,no quantitative indicators or timeframes in overall goals.Guidelines rarely promote community involvement,viewing bird-friendly green spaces more as landscape projects than as conservation efforts.At the district scale,focus is on hubs,corridors,and stepping stones,but few address matrix planning,which is vital as ecological corridors depend on surrounding land,especially since birds move through the matrix daily.Planting trees and creating green spaces can improve permeability and reduce population isolation in the matrix.At the site scale,guidelines clearly cover site selection,surveys,habitat design,and monitoring,but are less specific on threat analysis,risk management,and post-implementation actions.Few address surrounding buildings,lighting,or construction management,which are crucial because they influence bird behavior and safety.Emphasizing habitat resources while neglecting artificial factors may increase collision risks as habitat quality improves.Variations also exist in detailed methods,such as target species identification—ranging from vague survey-based identification to detailed taxonomic analysis—affecting conservation efforts and evaluations.Most guidelines focus on bird conservation;only one also considers ecological and visual value and human-bird interactions,which could foster local support and sustainability.Based on the above analysis,future efforts in developing new guidelines or revising existing ones should focus on:1)Risk management—reducing bird collisions,controlling stray cats,regulating lighting and feeding behaviors to prevent ecological traps;2)minimizing construction impacts on native species;3)engaging communities throughout all phases—from species selection and design to monitoring and enforcement—to ensure long-term conservation involvement. [Conclusion]The fundamental principles and content checklist outlined in this research are derived from a synthesis of existing knowledge,primarily intended to serve as a reference for developing guidelines related to bird-friendly green spaces.They provide a structured framework to support the development of guidelines for bird-friendly urban green space,thereby advancing the field.Future research may include monitoring and evaluation data from post-implementation assessments of bird-friendly green spaces across different regions to continually update and refine the principles and checklist presented in this study.

BackgroundAmbient air pollution is a potential environmental risk factor for inflammatory bowel disease (IBD), whereas a plant-based diet may be protective. However, whether such a dietary pattern modifies the association between long-term air pollution exposure and IBD risk remains unclear.MethodsThis prospective cohort study included 165,722 participants from the UK Biobank. Annual average concentrations of ambient air pollutants were estimated using air dispersion models. The plant-based diet index (PDI) was calculated based on the intake of 17 major food groups recorded in the 24-h dietary recall. Time-dependent Cox regression was applied to examine the association between long-term air pollution exposure and IBD risk, and stratified analyses along with interaction terms were performed to evaluate potential modification by the PDI.ResultsThis study demonstrated a positive association between air pollution exposure and the incidence of IBD, regardless of pollutant type. Air pollution exposure was positively associated with IBD risk among participants with lower PDI, whereas the point estimates were smaller and no longer statistically significant among those with higher PDI. A similar pattern was observed for ulcerative colitis.ConclusionLong-term exposure to ambient air pollution may be associated with an increased risk of IBD. In the higher plant-based diet group, air pollution-related IBD risk was relatively lower, providing additional epidemiological evidence into the prevention and management of IBD.

In natural language processing, misclassification of highly similar documents belonging to different categories is a longstanding challenge. Existing models struggle to effectively capture both global and local features. Furthermore, the presence of high-frequency irrelevant words and low-frequency yet semantically significant words further degrades the quality of document representations. To address these challenges, we propose a novel dual-level cloud-fuzzy relational graph contrastive learning network that introduces a granularity-aware approach to modeling document semantics. Distinct from traditional fuzzy methods, we incorporate the cloud model to meticulously characterize the inherent uncertainty-specifically the fuzziness and randomness-within document-word interactions. Our method captures 2nd-order fuzzy relationships among documents and words by constructing a document-word-document (DWD) graph. At the local level, we decompose each document into hierarchical units and build a 1st-order fuzzy adjacency graph using word-word, word-document, and document-word membership functions. To capture broader thematic similarities between documents, we extend this structure to a 2nd-order fuzzy graph, where document-document edges are inferred via fuzzy conjunction operations based on shared word anchors. Finally, graph contrastive learning takes as the input joint representation calculated from different views and hierarchies. Experimental results on several benchmark datasets demonstrate the advantages of DWD over state-of-the-art methods.

Urban green space has a significant impact on the urban quality of life and social benefits. Still, current methods for assessing its impact often overlook critical factors such as accessibility and subjective perception, thereby limiting the ability to capture the complex interactions between residents and urban green spaces. In this study, we have developed a new indicator, Perceived Green Volume (PGV), to measure urban residents' perceptible exposure to green space in their living environments. The indicator was constructed using high-resolution vegetation land cover data, perception-based structural coefficients of different vegetation types, and physical accessibility of green space types. We applied PGV to quantify green exposure in Beijing, China, and compared it with conventional indicators. Additionally, we estimated changes in PGV under different scenarios for the accessibility of urban green spaces. The results showed significant differences between traditional methods and PGV, with the latter providing a more relevant understanding of green exposure. Furthermore, the PGV framework can inherently support scenario-based analyses, enabling quantitative "what-if" evaluations of planning interventions, for example that opening access to underutilized green spaces in gated companies and institutions on non-working days can improve PGV up to 26.3 % in areas with limited accessibility. Our findings highlighted that by linking the qualitative perception of green spaces with its quantitative spatial availability, the PGV framework could enable a more human-centered evaluation of urban green exposure, thereby improving the understanding of urban residents' interactions with urban green spaces.

Green spaces offer a passive yet impactful component of urban design, with the potential to enhance cognitive function and reduce the risk of dementia in older adults. While the cognitive benefits of green spaces are recognized, particularly in improving cognitive performance among older populations, this area remains a critical yet underexplored domain within urban health research. This study aimed to investigate the association of exposure to green spaces with cognitive function among older adults across urban Beijing. A total of 62,333 participants from 114 communities were included in a city-wide screening program, utilizing an adapted version of the Mini-Mental State Examination (MMSE) and the Episodic Memory Test (EMT) to assess cognitive function. Green space exposure was quantified at the community level through multiple metrics, including accessibility (distance to large green spaces and parks), availability (Normalized Difference Vegetation Index [NDVI] and proportion of green space), and visibility (green view ratio). Multilevel linear and logistic regression models, adjusted for demographic and environmental covariates, were applied to data from 324 community healthcare centers. The findings indicated that greater exposure to green spaces, across all measured metrics, was associated with better cognitive function scores. A significant threshold was identified at 500 m from parks, beyond which cognitive function scores declined progressively with increasing distance. These results have important implications for urban planning, suggesting the need for accessible, community-level green spaces to support cognitive health for older adults in urban China.

Although the health benefits of greener environments are well documented, longitudinal evidence linking multidimensional greenness to blood lipid health, a key risk factor for cardiovascular diseases, remains limited, particularly in high-density urban settings. This 10-year (2008-2017) cohort study investigated the associations between multidimensional greenness and lipid health in Hong Kong. We assessed aerial greenness, street-level visibility (Green View Index), and three-dimensional greenness volume at each participant's addresses. We examined the associations between greenness and blood lipid profiles (total cholesterol, triglycerides, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol) and the development of dyslipidaemia among 24,458 adults and 6,239 adults, respectively. Using time-varying Cox regression models and linear mixed-effects models, we found that each interquartile range increase in all greenness metrics was associated with 8-14% reduced risk of the development of dyslipidaemia. All greenness metrics were also associated with healthier levels of lipid profiles. Notably, we identified potential thresholds (1.1 m3) for the health impacts of three-dimensional greenness volume, which require an increase of approximately 37.5% above current average levels in Hong Kong. Mediation analyses revealed that fine particulate matter significantly mediated up to 40.39% of the impact of three-dimensional greenness volume on the development of dyslipidaemia. Non-elderly adults and female participants appeared to derive smaller health benefits from greenness. These findings provide quantitative targets for urban greening, and also a mechanistic justification for using nature-based solutions to build a sustainable healthy city.

Urbanization is affecting biodiversity globally. Biotic homogenization is often cited as a key consequence. However, our understanding of this phenomenon may be biased by flaws in the methods used to document it. Here we estimate compositional dissimilarity among 39 urban tree assemblages worldwide while controlling for differences among regional species pools. Our results demonstrated the absence of a distinct global pattern in urban tree homogenization or differentiation. Homogenization mainly occurred among urban tree assemblages across broad geographic distances, whereas differentiation occurred at short distances. Nonnative species were a major contributing factor to these patterns. Sharing different nonnative species contributed to differentiation at short distances, whereas sharing the same nonnative species contributed to homogenization at broad distances. Our findings reveal a scale-dependent effect of urbanization on urban tree assemblages driven by nonnative species, emphasizing the global influence of urbanization on spatial patterns of biodiversity.

Objectives: To quantify the global burden of occupational noise-induced hearing loss (ONHL) from 1990 to 2021 by sex, age, and Socio-demographic Index (SDI), and to project its future trends through 2035. Methods: Data on disability-adjusted life years (DALYs), age-standardized DALY rates (ASDRs), and summary exposure values (SEVs) of ONHL were extracted from the GBD 2021 study. Joinpoint regression was used to estimate annual percent changes in ASDRs. An age-period-cohort model assessed the effects of age, period, and cohort. Autoregressive integrated moving average (ARIMA) models forecasted DALY trends from 2025 to 2035. Results: Globally, ONHL-related DALYs increased from 3.84 million in 1990 to 7.85 million in 2021, with the highest relative increases in low- and middle-SDI regions. Males consistently had higher burden than females. DALYs peaked in midlife (ages 50-59), while age-specific DALY rates peaked at older ages (60-74). APC analysis revealed significant age effects across all SDI levels, while period and cohort effects were limited. ARIMA models predicted that global ONHL DALYs would reach 9.65 million by 2035, a 22.97% increase from 2021, with largest increases expected in those aged 55-59. Conclusions: Occupational noise–induced hearing loss remains a growing global public health challenge, particularly among male workers in lower-SDI regions. Continued increases are projected over the next decade. These results underscore the urgent need for strengthened hearing conservation strategies and targeted prevention policies to reduce ONHL burden worldwide.

BACKGROUND:Intensifying heatwaves driven by climate change pose a severe and disproportionate health burden, particularly for populations undergoing rapid demographic and environmental transitions, yet globally comparable projections of future mortality under varying scenarios remain limited. METHODS:We constructed a global, grid-based, and three-stage framework covering 176 countries, integrating climate projections from 20 CMIP6 models under four Shared Socioeconomic Pathways (SSP1-2.6, SSP2-4.5, SSP3-7.0, and SSP5-8.5). Heatwaves were defined as ≥2 consecutive days with grid-specific daily mean temperatures exceeding the 95th percentile. Historical heatwave-mortality associations from 750 locations, together with future population distributions, urbanisation trajectories, and adaptation scenarios, were incorporated to estimate excess heatwave-attributable deaths during the 2030s, 2050s, and 2090s, with uncertainty quantified via Monte Carlo simulations. FINDINGS:Global heatwave-related deaths are projected to increase from 1,018,776 (95% UI: 935,188-1,106,785) in the 2000s to 5,289,925 (95% UI: 2,665,816-8,287,210), 8,814,669 (95% UI: 5,055,306-13,734,514), 13,690,238 (95% UI: 9,902,215-19,250,977), and 16,133,330 (95% UI: 11,559,920-21,300,978) in the 2090s under SSP1-2.6, SSP2-4.5, SSP3-7.0, and SSP5-8.5 scenarios, respectively. Sub-Saharan Africa will suffer from the largest heatwave-related deaths in the 2090s, contributing to the proportion of approximately 20.33%-24.92%. Projected slopes are steeper under low urbanisation level, SSP3 population scenario, and no adaptation scenario. INTERPRETATION:The projected heatwave mortality burden underscores the urgent need for targeted climate adaptation, urban planning, and public health strategies. The findings reveal differential vulnerability across populations and regions, offering guidance for resource allocation to strengthen resilience in a warming world. FUNDING:National Natural Science Foundation of China (No. 42575195), China Postdoctoral Science Foundation (No. 2025M780707), and Guangdong Provincial General Colleges and Universities Innovation Team Project (Natural Science) (No. 2024KCXTD004).

Urban forests are critical for climate adaptation and liveability, but effective irrigation management-key to their sustainability-remains poorly documented at the global scale. This study addresses this critical knowledge gap by analysing urban forest irrigation practices across 109 cities in 21 countries, offering one of the first global assessments of irrigation approaches, challenges, and opportunities. Using survey data, we examined water sources, irrigation frequency, constraints, and enabling conditions. Our results show that weather conditions were the leading factor influencing irrigation scheduling in 44 % of cities, while 56 % reported no formal water restrictions. Despite the importance of water conservation, 55 % of respondents reported having no water usage monitoring systems, and 73 % lacked financial incentives to promote water-efficient irrigation. A large majority (80 %) did not use recycled wastewater, and 58 % did not conduct water quality testing. Only 15 % of cities regularly used water-efficient irrigation technologies, and 47 % had no plans to implement smart systems. Over half (56 %) rated their current irrigation practices as only moderately successful. Budget constraints and infrastructure limitations were the most frequently reported challenges, followed by climate change-related concerns. While environmental variables such as mean annual temperature and irrigation need influenced specific practices, local governance and institutional actions had stronger effects. Cities in the Global South reported distinct strategies and constraints compared to those in the Global North. Our findings provide actionable insights for climate-resilient urban water strategies and underscore the need for targeted policies, capacity-building, and efficient technologies to enhance urban forest sustainability worldwide.

Despite extensive research, air pollution-influenza associations remain inconsistent. This evidence synthesis aggregated evidence from 28 studies to quantify air pollution-influenza associations, employing the Office of Health Assessment and Translation (OHAT) and Navigation Guide tools for quality assessment. We derived combined relative risk (RR) per 10 µg/m3 increase in air pollution, 95% confidence intervals (CIs) and 95% prediction intervals (PIs) to quantify the link. PM2.5 (RR = 1.037; 95% CI: 1.019-1.055), SO2 (RR = 1.351; 95% CI: 1.107-1.649) and CO (RR = 1.003, 95% CI: 1.001-1.004) showed significant links to higher influenza risk. PM10 (RR = 1.064, 95% CI: 0.964-1.174), NO2 (RR = 1.207, 95% CI: 0.988-1.475), and O3 (RR = 1.027, 95% CI: 0.962-1.097) demonstrated a nonsignificant tendency to increase influenza risk. Stratified analyses indicated heightened vulnerability to SO2 in Asian populations relative to Australians, and to PM2.5 in Europe compared to Asia. Males experienced greater risks for PM2.5, PM10, NO2 and SO2. However, these associations were characterized by substantial between-study heterogeneity even after stratification, which was reflected in 95% PI that suggested an inconsistent direction of effect in future studies. Our analysis confirmed air pollutants increase influenza risk, but confounding, exposure variability, and mechanistic pathways warrant further study, particularly in underrepresented regions.

The integration of remote sensing technology and deep learning has significantly advanced object recognition in high-resolution satellite imagery. However, most existing datasets mainly focus on stumpy objects, while slender targets, characterized by a height that significantly exceeds their length or width, have been overlooked. This gap highlights the need for specialized datasets to improve recognition of slender objects. In response, this study developed the Transmission Tower Oriented Bounding Box (TT-OBB) dataset, which systematically classifies transmission towers as representative slender targets into four distinct categories. To validate the effectiveness of deep learning methods in recognizing these structures, comprehensive experiments were conducted using several mainstream detectors on the TT-OBB dataset. The results demonstrate the practicality of this dataset, highlighting its utility as a reference for deep learning applications in transmission tower identification, and providing valuable insights for model selection in real world scenarios

Local climate zone (LCZ) provides a detailed classification system for building types in urban areas and offers a unified standard for block-scale surface urban heat island (SUHI) studies. However, LCZ mapping methods with high classification accuracy for global applicability, and multi-city comparison of SUHI based on LCZ are still needed. In this study, we developed a transferable LCZ mapping framework for 30 China cities at 120 m by using multi-source remote sensing and GIS data, and random forest model. The gap-filled LST for 21 cities among 30 cities with diverse urbanized levels and climate conditions were generated, utilizing Landsat 8 data, LST retrieval algorithm and gap-filling method. Spatial patterns of SUHI intensity across climate zones and cities of different sizes were explored, and the impacts of urban morphology on SUHI in built-up LCZs were analyzed using the boosted regression trees model. Results showed that the proposed LCZ mapping framework achieved high accuracy in China cities, with overall accuracy from 0.86 to 0.93. Its robustness and transferability were further demonstrated in three cities in the United States with overall accuracy of 0.91 to 0.93. LST gap-filling method also performed well, with R from 0.71 to 0.91 and RMSE from 1.86 degrees C to 3.59 degrees C, respectively. Our multi-city assessment revealed consistent patterns of SUHI in LCZs across climate zones: compact mid-rise (LCZ 2) and large low-rise (LCZ 8) had the highest SUHI intensity, while sparsely built (LCZ 9) and open low-rise (LCZ 6) had the lowest values. Moreover, LCZ 2 tended to have higher SUHI intensity in colder climate regions, while LCZ 8 exhibited higher values in warmer climate regions. City size also influenced SUHI effect in built-up LCZs, with large cities exhibiting SUHI intensity up to 1 degrees C higher than small cities. Additionally, vegetation exhibited the largest of relative importance (20 % to 68 %) which impacted SUHI intensity in built-up LCZs, with a higher value in cold cities compared to warm cities. Impervious or building surface fraction also accounted for 13 % to 45 % of the SUHI contribution across LCZs, with relative importance about 3 % to 10 % greater in warmer and larger cities. The findings of this study can be useful in developing urban planning policies for intra-city SUHI mitigation, and our transferable LCZ mapping framework can be applied to other global cities for SUHI studies.