This study aims to develop and evaluate a machine learning (ML)-based visibility forecasting method for the winter-spring months (January, February, and March) at Noi Bai International Airport, Vietnam. We developed tree-based machine learning models, including Decision Tree (DT), Random Forest (RF), and XGBoost (XGB), using surface meteorological variables and Global Forecast System (GFS) data to forecast visibility at 1-, 2-, and 3-hour lead times. The results indicate that many predictors did not improve forecasting performance. In contrast, DT models with just 3–5 key predictors (selected via RFECV and SHAP) achieved optimal results, particularly in detecting low- and rare-visibility events. The most influential predictors included lagged visibility, time of day, and the vertical gradient of relative humidity within the near-surface and atmospheric boundary layers. The optimal DT models consistently achieved skill scores for the Probability of Detection (POD) of low- and rare-visibility classes ranging from +10% to +35% compared to the persistence forecast, while operating stably without evidence of overfitting.
Organic carbon in precipitation represents an important pathway linking atmospheric processes with terrestrial and aquatic carbon cycling, yet data from rapidly urbanizing tropical and subtropical cities remain scarce. This study investigated total organic carbon (TOC) concentrations in 264 rainwater samples collected from 21 wards across Hanoi, Vietnam, between May 2023 and December 2025. TOC concentrations ranged from 0.41 to 17.72 mgC.L-1, with an average of 3.75±2.94 mgC.L-1, exceeding mean values previously reported for Asia and global continental regions. Significant seasonal variability was observed, with TOC concentrations consistently higher during the dry season than the rainy season, reflecting the dominant influence of rainfall dilution and atmospheric washout processes. The study also detected an upward trend in annual mean TOC concentrations over the study period, suggesting increasing contributions from anthropogenic emissions. Despite Hanoi receiving relatively high annual rainfall, elevated TOC concentrations indicate that urban activities remain an important source of atmospheric organic carbon available for wet deposition. This study provides critical baseline data to improve regional and global estimates of wet organic carbon deposition, advance understanding of carbon cycling in urban environments, and support strategies for air quality management and sustainable rainwater use.
Tropical montane forests in the mainland Southeast Asia biodiversity hotspot are increasingly exposed to climate warming and recurrent drought, yet long-term evidence of how these changes affect endangered tree species remains limited. Here, we developed multi-century ring-width (RW) and basal area increment (BAI) chronologies of Fokienia hodginsii (Cupressaceae), a long-lived conifer from Pù Mát National Park in the northern Annamite Mountains of Vietnam, to investigate long-term growth dynamics and climate responses. Individual trees showed contrasting growth trajectories, including early release, gradual ascent, and suppressed release patterns. In contrast, RW and BAI chronologies shared a common signal among the sampled dominant trees over more than three centuries. Climate-growth analyses showed that pre-monsoon moisture availability during March-May dominated growth, with both RW and BAI positively associated with precipitation and the Palmer Drought Severity Index (PDSI) and negatively associated with temperature. Spatial correlations further indicated that these relationships reflect broader regional hydroclimatic variability across mainland Southeast Asia. Comparison with independent drought records showed that major drought episodes during the mid-eighteenth and late-nineteenth centuries coincided with pronounced growth reductions, indicating that drought has constrained F. hodginsii growth from interannual to multidecadal timescales. Both chronologies also show reduced growth in recent decades, which may reflect increasing moisture stress under contemporary warming. These results indicate persistent drought sensitivity in the sampled dominant F. hodginsii trees over the past three centuries, suggesting that intensifying moisture stress under warming could threaten the species' long-term persistence at the site, although the small sample (nine trees from one ridge) limits generalization to the whole population. We propose monitoring seedling recruitment failure and canopy mortality on shallow, well-drained ridge-top soils as priority early-warning indicators to guide adaptive forest management in Vietnam's protected areas.
The Nalsarovar Wetland Basin, a Ramsar-designated wetland covering approximately 120 km² in the semi-arid region of Gujarat, India, represents a distinctive Quaternary wetland system whose evolution has been shaped by interactions among tectonic activity, sedimentation, hydrogeochemical processes, climatic variability, and increasing anthropogenic pressures. This study presents an integrated geoenvironmental synthesis based on published geological, geomorphological, sedimentological, hydrogeochemical, ecological, environmental, and remote-sensing investigations to reconstruct the basin's geological evolution and evaluate its contemporary environmental transformation. Geological evidence indicates that the basin evolved from a shallow marine or estuarine environment within the tectono-geomorphic Nal Corridor into an inland wetland during the Late Quaternary and Holocene, driven by marine regression, sediment accumulation, and hydrological isolation. Remote-sensing datasets derived from IRS-P6/Resourcesat LISS-III imagery (23.5 m spatial resolution) achieved overall classification accuracies of 91.46% (Kappa = 0.9004) for pre-monsoon and 87.91% (Kappa = 0.8621) for post-monsoon conditions, confirming significant changes in wetland extent, vegetation distribution, saline surfaces, and surrounding land use. The integrated synthesis demonstrates that accelerated sedimentation, basin infilling, groundwater salinization, nutrient enrichment, and anthropogenic activities, including agricultural expansion, groundwater extraction, and land-use modification, are progressively transforming the wetland into a more saline, sediment-filled, and partially terrestrialized ecosystem. The study shows that current environmental conditions in the Nalsarovar Wetland are shaped by the interaction between inherited geological factors and increasing anthropogenic pressures. The integrated geoenvironmental framework developed in this study provides a scientific basis for sustainable wetland management, long-term environmental monitoring, and conservation of semi-arid Ramsar wetlands under changing climatic and land-use conditions, while identifying high-resolution geospatial monitoring and long-term hydrogeochemical investigations as important priorities for future research.
Rapid urbanization in large cities has posed many environmental challenges, especially air pollution. This study used multi-source remote sensing data to analyze the impact of urbanization on air pollution in Hanoi during 2021–2025. Landsat 8/9 Collection 2 Level 2 Surface Reflectance imagery was used to map urbanization through the Index-based Built-up Index (IBI), combined with land-use classification based on machine learning algorithms, Random Forest (RF) and Support Vector Machine (SVM), calibrated with 331 field-collected samples. A total of 99 Landsat scenes (single-date summer images for RF/SVM classification and multi-temporal dry-season composites for IBI analysis) were processed across the study period. We characterized air pollution using daily MODIS MCD19A2 AOD products (1 km resolution), validated against 55 Aerosol Robotic Network (AERONET) Level 2.0 measurements at Nghia Do station (R² = 0.62) and correlated with hourly PM₂.₅ concentrations from 11 ground monitoring stations (AQMIS and OpenAQ networks). We developed pollution distribution maps for each year and used statistical tests (Root Mean Square Error (RMSE), R²) to assess the reliability of the satellite data. The study spatially overlaid urbanization and air pollution maps, ran correlation analyses using Moran's I, and built univariate and multivariate regression models incorporating ERA5 meteorological covariates to quantify the relationship between urbanization and pollutant concentrations. The results showed that built-up area (IBI > 0) expanded from 417.77 km² (12.4%) in 2021 to 701.36 km² (20.9%) in 2025. The linear regression between RF-classified urban area and mean AOD yielded R² = 0.82 (p < 0.001), while AOD correlated with ground-level PM₂.₅ at R² = 0.34–0.65 across 11 stations (improved to R² = 0.48–0.80 with meteorological covariates). Spatial autocorrelation analysis confirmed significant clustering of both urbanization (Moran's I = 0.62) and pollution (Moran's I = 0.78). The study highlighted the role of satellite data in urban environmental monitoring and provided a scientific basis for sustainable urban development planning.
