Application of Machine Learning and Remote Sensing in Hydrology

Water is vital to all life on earth, but its management is becoming more difficult owing to the behavior of water in nature such as water dynamics, water movements, and different forms of water in nature. In addition, population growth, the impact of climate change, and the inappropriate use of water resources add more complexity to water resource management. The integration of natural sciences necessitates creative methods in decision sciences, data processing, and modeling methodologies. The prediction of such occurrences is a highly nonlinear problem that requires the use of modern capable techniques. For handling the abovementioned items, we need to use tools capable of solving water-based issues. Machine learning and remote sensing technologies as recently developed technologies have been considered as tools for these solving water-based issues. Additionally, the “Topic” sought to obtain a collection ofrecent studies on the application of machine learning and remote sensing in hydrology.

Overview of the Topic

Metwaly et al. [1] presented a hydrogeophysical study of the sub-basaltic Alluvial aquifer in the southern part of Al-Madinah Al-Munawarah (Saudi Arabia). Their findings suggested that surface geoelectrical resistivity approaches may offer an alternate, quick, and cost-effective way for calculating aquifer hydraulic parameters in cases when pumping data are rare or unavailable.

Lee et al. [2] researched aquatic ecosystem health index predictions using machine learning approaches. They applied the Wasserstein generative adversarial network, and their findings demonstrated that the used machine learning model’s performance was satisfactory for aquatic ecosystem health studies.

Research on the identification of dominant factors in the groundwater recharge process based on multivariate statistical approaches was studied by Castillo et al. [3]. This research indicated that the Sierra de San Miguelito Volcanic Complex (SSMVC) does not function well as a water recharge zone towards the deep aquifer of the San Luis Potos Valley (SLPV) due to its climate and geology. This technique will aid water resource managers in identifying and defining recharge regions with better precision.

The paper by Yang et al. [4] studied soil moisture retrieval by microwave remote sensing data and a deep belief network (DBN) in the Tibetan Plateau. The DBN soil moisture model performed well with a ten-fold cross-validation strategy. The accuracy of the prediction was enhanced when the bare-soil backscatter coefficient was utilized as training data.

Liu et al. [5] predicted soil moisture via a backpropagation neural network model optimized with a genetic algorithm (GA-BP). They applied the backpropagation neural network and the GA-BP models for soil moisture prediction with and without lag in two different conditions, and the results showed that considering lag may improve the accuracy of prediction compared with predictions without considering lag.

The paper by Wang et al. [6] showed that the applied novel intelligent inversion approach is dependable for swiftly and precisely acquiring hydrogeological parameters, serving as a benchmark for the inversion of parameters in other fields.

Jiang et al. [7] used a multi-model strategy based on random forest and Sentinel-2 images for water information extraction. The findings indicate that enhanced normalized difference water index (MNDWI), band B2 (Blue), normalized water index (NDWI), B4 (Red), B3 (Green), and band B5 (Vegetation Red-Edge 1) have significant influences on the accuracy of the model.

Islam et al. [8] assessed the impact of the Farakka Barrage on hydrological alteration. According to the results of their continuous wavelet analysis, the Farakka Barrage has a substantial effect on the periodicity of the streamflow regime. This paper discussed the measurement of hydrological change and estimated the discharge for forthcoming days, which may be useful for the development of sustainable water resource management strategies.

The author expects that all readers of this “Topic” will benefit from the machine learning and remote sensing techniques applied in these water resources management issues.