Search

Groundwater resources are required for domestic water supply, agriculture, and industry, and the strategic importance of water resources will only increase in the context of climate change and population growth. For optimal management of this crucial resource, exploration of the potential of groundwater is necessary. To this end, the objective of this study was the development of a new method based on remote sensing, deep neural networks (DNNs), and the optimization algorithms Adam, Flower Pollination Algorithm (FPA), Artificial Ecosystem-based Optimization (AEO), Pathfinder Algorithm (PFA), African Vultures Optimization Algorithm (AVOA), and Whale Optimization Algorithm (WOA) to predict groundwater potential in the North Central region of Vietnam. 95 springs or wells with 13 conditioning factors were used as input data to the machine learning model to find the statistical relationships between the presence and nonpresence of groundwater and the conditioning factors. Statistical indices, namely root mean square error (RMSE), area under curve (AUC), accuracy, kappa (K) and coefficient of determination (R2), were used to validate the models. The results indicated that all the proposed models were effective in predicting groundwater potential, with AUC values of more than 0.95. Among the proposed models, the DNN-AVOA model was more effective than the other models, with an AUC value of 0.97 and an RMSE of 0.22. This was followed by DNN-PFA (AUC=0.97, RMSE=0.22), DNN-FPA (AUC=0.97, RMSE=0.24), DNN-AEO (AUC=0.96, RMSE=0.25), DNN-Adam (AUC=0.97, RMSE=0.28), and DNN-WOA (AUC=0.95, RMSE=0.3). In addition, according to the groundwater potential map, about 25–30% of the region was in the high and very high potential groundwater zone; 5–10% was in the moderate zone, and 60–70% was low or very low. The results of this study can be used in the management of water resources in general and the location of appropriate wells in particular. [ABSTRACT FROM AUTHOR]

Seagrasses are among the most important ecosystems of continental and island coastal waters globally. Analyses of the species diversity, structure, and distribution mapping of seagrass communities provide a scientific basis for seagrass conservation and development and protection of the marine environment. The application of remote sensing, especially using very high-resolution remote sensing data, in mapping seagrass distribution confers high accuracy and detailed results. The distribution of seagrass communities and benthic biomes typical of Nam Yet Island, Truong Sa Archipelago, Vietnam was mapped using object-based classification of Pléiades satellite imagery with four spectral bands (red, blue, green, and near-infrared) and the Green–Blue Normalized Difference Vegetation Index, combined with handheld positioning devices (total field samples 100, overall accuracy 91%, kappa coefficient = 0.88). Besides, by sampling and measurement in the field, two seagrass species, Thalassia hemprichii and Halophila ovalis were identified with the same structure, density, and biomass in the northern and southern parts of the island. The application of high-resolution remote sensing images combined with field surveys contributes to improving the accuracy of seagrass distribution maps as well as benthic biome ones for shallow sea areas. Establishing a seagrass distribution map and assessing the diversity of seagrass is the basis for the conservation and development of the current seagrass community at Nam Yet Island. [ABSTRACT FROM AUTHOR]