The Mekong Basin is the most critical transboundary river basin in Asia. This basin provides an abundant source of fresh water essential for the development of agriculture, domestic consumption, and industry, as well as for the production of hydroelectricity, and it also contributes to ensuring food security worldwide. This region is often subject to floods that cause significant damage to human life, society, and the economy. However, flood risk management challenges in this region are increasingly substantial due to conflicting objectives between several countries and data sharing. This study integrates deep learning with optimization algorithms, namely Grasshopper Optimisation Algorithm (GOA), Adam and Stochastic Gradient Descent (SGD), and open-source datasets to identify the region of probably occurring floods in the Mekong basin, covering Vietnam and Cambodia. Various statistical indices, namely Area Under the Curve (AUC), root mean square error (RMSE), mean absolute error (MAE), and coefficient of determination (R²), were used to evaluate flood susceptibility models. The results show that the proposed models performed well with AUC values above 0.8, specifying that the DNN-Adam model achieved an AUC of 0.98, outperforming DNN-GOA (AUC = 0.89), DNN-SGD (AUC = 0.87), and XGB (AUC = 0.82. Regions with very high flood susceptibility are concentrated in the Mekong Delta of Vietnam and along the Mekong River in Cambodia. The findings of this study are significant in supporting decision-makers or planners in proposing appropriate flood mitigation strategies, planning policies, and strategies, particularly in the Mekong River watershed.

This study investigates the spatio-temporal patterns of drought and their teleconnection with land surface properties in Yok Don National Park during the dry season using the Temperature-Soil Moisture Dryness Index (TMDI). This index is derived from the relationship between the Normalized Difference Land Heat Index and Land Surface Temperature, extracted from Landsat-8 data acquired in the mid-dry season from 2014 to 2023. Results reveal increasing drought severity starting in 2014, peaking during 2015–2016, and decreasing from 2017 to 2020. Drought conditions escalated again during 2021–2022 before moderating by 2023. These trends align with in-situ precipitation data recorded at a nearby meteorological station, highlighting varied impacts on forest types. Areas covered by deciduous broadleaf forests experienced pronounced drought effects, whereas evergreen broadleaf forests showed greater resilience. Land surface evapotranspiration rates obtained from NASA’s MOD16A2GF dataset were used to evaluate TMDI performance. During the dry seasons from 2014 to 2023, TMDI exhibited a consistent negative correlation with evapotranspiration, with coefficients ranging from -0.55 to -0.70. This demonstrates TMDI's effectiveness in capturing land surface water availability and assessing drought conditions. The findings provide crucial insights into drought monitoring and management for Yok Don National Park and other water-scarce regions, reinforcing TMDI’s value in sustainable forest management and drought mitigation.

The stable downward continuation method can enhance low amplitude anomalies and improve the resolution of potential fields. However, this method may be affected by the edge effect. In this paper, we improve the downward continuation results by using the cubic Hermite interpolation to extend data. Our synthetic model shows that the proposed extension can provide better estimates, especially at the edges, than using constant values. We further enhance downward continued data using boundary filters where the boundaries obtained from the downward continued data have a higher resolution. The methods are also applied to interpret RTP magnetic and Bouguer gravity anomalies of the Central Highlands (Vietnam), where the RTP magnetic data are determined from the multiple-stage RTP method, while the Bouguer gravity data are calculated using the Parker formula. The spectral analysis of potential fields reveals the depth of shallow sources in the area of about 1 km, which determines the height of the downward continuation. A new subsurface structural map is established for the Central Highlands from the filtered results of downward continued data, which will be a helpful document for detailed future explorations and geology studies of the area.

Highly effective adsorbents derived from modified Di Linh bentonite (Lam Dong Province, Vietnam) were produced using Al3+ stock solution prepared from Al2(SO4)3•18H2O. Mineral, morphology, and surface area properties of untreated and Al-modified Di Linh bentonite were characterized using X-ray diffraction, Fourier transform infrared spectroscopy, scanning electron microscopy, and nitrogen adsorption-desorption analyses. A scanning experiment was conducted to investigate the Al-modified bentonite material's ammonium (NH₄⁺) removal capacity at two levels of NH₄⁺ initial concentration. Results show that the combination of acid H2SO4 (a bioproduct of diluted Al solution) with Al3+ caused the smectitization of clay particles via a dissolution-precipitation mechanism, which enhances the structural organization of smectite and modifies its mineralogical properties. This process promoted the removal capacity of Al-modified bentonites, which increased to 0.47 mg/g in comparison with 0.32 mg/g from untreated bentonite at 50 mg/L NH4+-N concentration, to 19.3 mg/g in comparison with 17.2 mg/g from untreated bentonite at 1000 mg/L NH4+-N concentration. This approach to modifying natural bentonite offers new possibilities for developing adsorbents to eliminate NH4+ from water.

The precision of estimating soil salinity is considered a key task in solving soil salinity problems and irrigation management of agriculture. This problem is increasingly important in the Mekong Delta, where it is severely affected by this phenomenon in the context of climate variability. Therefore, this paper aims to construct a soil salinity map with high accuracy using machine learning and Sentinel 2A, namely Xgboost (XGB) and Random Forest (RF). The province of Tra Vinh in the Mekong Delta has been selected as the case study. 68 soil salinity samples were collected in August 2024, and 25 conditioning factors extracted from the Sentinel 2A image were used as input data for the machine-learning model. Three statistical indices, namely root mean square error (RMSE), mean absolute error (MAE), and coefficient of determination (R2), were used to evaluate the effectiveness of machine learning models. The results showed that with an R2 value of 0.86, the XGB model was superior to the RF model with an R2 value of 0.67. Furthermore, Tra Vinh province, the coastal region, and along the Mekong River are more severely affected by soil salinity with an electrical conductivity (EC) value of more than 10. This region, more affected by soil salinity, is related to rising tides and sea levels in the context of climate variability. This study plays an important role and can support farmers in regions affected by soil salinity in building investment measures to reduce the impacts of soil salinity on the development of agriculture.

Slope instability is a common geological hazard along mountainous roads in Vietnam, leading to significant damage to infrastructure, traffic disruptions, and loss of life. Predicting slope stability, typically quantified by the Factor of Safety (FS), is challenging due to the complex interactions between geotechnical, topographical, and environmental factors. This study aims to develop efficient and accurate models for predicting the FS of natural slopes using advanced machine learning techniques, including Gradient Boosting (GB), Support Vector Machine (SVM), Multi-layer Perceptron (MLP) Neural Networks, Random Forest (RF), and AdaBoost (AB). 371 slope stability cases were used to create a comprehensive database for model training. Both geotechnical and topographical parameters were considered in the FS prediction process. The performance and reliability of these models were evaluated using standard metrics such as R², MAE, and MSE. The results demonstrated that all models exhibited satisfactory prediction capabilities, with the optimized GB model achieving the highest accuracy (R² = 0.975, MAE = 0.079, and RMSE = 0.120). Additionally, SHAP analysis was employed to assess the importance of input variables in predicting the FS. The findings revealed that slope ratio (X1), slope height (X2), and the number of steeps (X3) were the most influential parameters in the FS prediction.

With the rapid advancement of technology, monitoring forest cover changes has become increasingly quantifiable through various techniques and methods. In this study, we developed a procedure that utilizes the Deep Neuron Network (DNN) model and the Geographic Information Systems (GIS) based on high-resolution imagery captured at different time points to create forest cover change maps in Nui Luot, Chuong My, Hanoi. Two RGB (Red-Green-Blue) spectral images were captured by Unmanned Aerial Vehicle (UAV) at two different time points (pre-scene and post-scene) and used to extract information for the DNN model to produce land cover maps for these two time points. The land cover classification was divided into four classes: (1) Trees, (2) Vacant, (3) Built area and others, and (4) Water surface. Combined with GIS analysis, the forest cover change maps were developed to quantify detailed increases or losses in forest cover based on the "Trees" class. The model's accuracy was evaluated using parameters such as the area Under the ROC Curve (AUC), Accuracy (ACC), Precision, Recall, F1-Score, Kappa, and Root Mean Square Error (RMSE). The analysis results indicate that from January 31, 2023, to October 20, 2023, the forest cover in the study area decreased by 0.53%. The accuracy metrics for the pre-change scene were: average AUC = 0.922, ACC = 76.86%, average Precision = 0.743, average Recall = 0.73, average F1-Score = 0.723, Kappa = 0.692, and RMSE = 0.297. For the post-change scene, the accuracy metrics were: average AUC = 0.954, ACC = 81.89%, average Precision = 0.823, average Recall = 0.815, average F1-Score = 0.818, Kappa = 0.758, and RMSE = 0.262. A deforestation scenario was constructed to evaluate the effectiveness of the DNN models in assessing and monitoring forest dynamics.

Drought is one of the most pervasive and complex natural hazards, significantly impacting ecosystems, agriculture, and communities, particularly in Vietnam. The study constructed a hybrid model to explore the sensitivity of drought forecast over Vietnam, utilizing bias-corrected precipitation and temperature data from regional climate models, RegCM, and clWRF. The resulting 6-month scale Standardized Precipitation Evapotranspiration Index (SPEI-6), is then processed through two different multi-model ensemble approaches: a simple averaging method (ENS) and a more complex artificial neural network (CTL), forming the basis of our two experimental setups. CTL consistently outperformed ENS, demonstrating more substantial drought-predictive skills. CTL effectively captured the spatio-temporal distribution of SPEI-6, showing high accuracy at a 1-month lead time. Its performance is promising, particularly in regions with complex climate patterns like the Central of Vietnam (R4 and R5), though discrepancies in predicting SPEI-6 amplitudes become slightly evident at a 5-month lead time. The geographic extent analysis further supports CTL's strengths in short-term forecasting, highlighting its utility in early warning systems and immediate drought response planning. Nonetheless, the decrease in accuracy at extended lead times underscores the need for model refinement. The study contributes to the growing body of literature on ANN-based drought forecasting, emphasizing the potential and limitations of these models in the context of Vietnam.

Flash floods continue to emerge as a serious and growing natural hazard for many communities worldwide, especially in areas affected by tropical storms. These floods damage critical infrastructure and severely strain economic resources, underscoring the urgent need for advanced flood prediction tools. This study presents an innovative integrated machine learning approach, BCMO-RF, which merges Balancing Composite Motion Optimization (BCMO) with Random Forest (RF) to map flash flood susceptibility. In the BCMO-RF approach, the RF algorithm is applied to develop the flash flood model, while BCMO is used to explore and optimize the model's parameters. The study concentrates on areas in Thanh Hoa Province, Vietnam, frequently impacted by flash floods. Accordingly, various geospatial data sources were utilized to compile a geodatabase comprising 2,540 flash flood locations and 12 influencing factors. The geodatabase served as the basis for training and validating the BCMO-RF model. Results show that the BCMO-RF model attained high prediction accuracy (93.7%), achieving a Kappa coefficient of 0.874 and an AUC score of 0.988, outperforming the Deep Learning model benchmark. The study finds that the BCMO-RF model is reliable for accurately mapping areas susceptible to flash floods.

Oligocene sediments in northern Vietnam have been extensively studied in terms of geology, stratigraphy, paleogeography, formation environments, tectonics, etc. However, relatively little attention has been paid to the paleoclimate. The Oligocene climate interpreted herein is based on features recorded in sediments taken from the Dong Ho (Hoanh Bo basin), Na Duong (Na Duong basin), and Co Phuc (Red River Trough) formations. These sediments were analyzed using thin-section microscopy, X-ray diffraction, and palynology with the Coexistence Approach. The sediments primarily consist of conglomerate, gritstone, sandstone, siltstone, claystone, and coal shale, deposited in continental environments and dated to the Oligocene based on palynomorph assemblages. The Oligocene paleoclimate is generally warm subtropical, with intermittent hot-humid or cold-dry periods and a slight influence of monsoons. An alternation of hot-humid and cold-dry climates was recorded in the Hoanh Bo basin. In the Red River Trough, the Oligocene climate exhibited a Mean Annual Temperature (MAT) of 9.3-22.2°C and a Mean Annual Precipitation (MAP) of 1122 to 1857 mm, based on the majority of palynological samples, indicating more continental, drier and colder conditions than present. In the Na Duong basin, a warm subtropical climate with a MAT of 9.3-21.7°C and MAP of 1122-1724 mm in the majority of samples was occasionally replaced by hot and humid subtropical periods; the variations in temperature and precipitation followed a similar pattern, suggesting an alternation between dry-cold and humid-hot phases.