The water cycle in the key agricultural and pastoral zones (KAPZs) is an important factor for maintaining the stability of the ecosystem. Groundwater collection and lateral seepage are indispensable parts of the water cycle, and it is difficult to monitor the groundwater situation in each area. The strength of the alternate circulation of groundwater is directly related to the utilization value and development prospects of groundwater; therefore, creating an effective method for the detection of groundwater burial depth has become an issue of increasing concern. In this paper, we attempt to create a method for the detection of groundwater burial depth that combines cokriging interpolation, spatial autocorrelation, geographically weighted regression, and other methods to construct a quantitative relationship between different land cover types and groundwater depth. By calculating the band index of the land cover type, the groundwater depthinformation of the unknown area can be obtained more accurately. Through collaborative kriging interpolation, normalized difference vegetation index (NDVI), precipitation, and hydrogeological conditions were used as covariates. The groundwater burial depth of Wengniute Banner in 2005, 2009, 2013, and 2017 was the response variable, and the groundwater burial depth in the study area was calculated. The groundwater burial depth data after the cokriging interpolation was used to transform the raster data into vector data in space using the improved hydrological response unit (HRU) model to make it more suitable for the actual groundwater confluence. Subsequently, 551 minimum response units (MHRUs) were obtained by division, and the spatial autocorrelation analysis was performed accordingly. The groundwater burial depth in the study area is spatially distinct from east to west, and the groundwater level shows a trend of being high in the west and low in the east, gradually increasing due to precipitation and rivers. The average change of groundwater depth in the time series is not significant, but it does gradually show a trend of accumulation. According to the aggregation characteristics of spatial autocorrelation analysis, a geographically weighted regression model of groundwater depth and NDVI, normalized difference drought index (NDDI), and net relatedness index (NRI) was established. The NDVI representing the forest land and the Adjusted R² of the groundwater depth is 0.67. The NRI representing the cultivated land and the Adjusted R² of the groundwater depth is 0.8675. The NDDI representing the bare land and the Adjusted R² of the groundwater depth is 0.7875. It shows that the band index representing the ground type has a good fitting effect with the groundwater burial depth. Full article

Accurate and timely landslide susceptibility mapping (LSM) is essential to effectively reduce the risk of landslide. In recent years, deep learning has been successfully applied to landslide susceptibility assessment due to the strong ability of fitting. However, in actual applications, the number of labeled samples is usually not sufficient for the training component. In this paper, a deep neural network model based on semi-supervised learning (SSL-DNN) for landslide susceptibility is proposed, which makes full use of a large number of spatial information (unlabeled data) with limited labeled data in the region to train the mode. Taking Jiaohe County in Jilin Province, China as an example, the landslide inventory from 2000 to 2017 was collected and 12 metrological, geographical, and human explanatory factors were compiled. Meanwhile, supervised models such as deep neural network (DNN), support vector machine (SVM), and logistic regression (LR) were implemented for comparison. Then, the landslide susceptibility was plotted and a series of evaluation tools such as class accuracy, predictive rate curves (AUC), and information gain ratio (IGR) were calculated to compare the prediction of models and factors. Experimental results indicate that the proposed SSL-DNN model (AUC = 0.898) outperformed all the comparison models. Therefore, semi-supervised deep learning could be considered as a potential approach for LSM. Full article

The over-extraction of groundwater from thin fresh groundwater lenses is a threat to the livelihood of farmers in the Lower Indus Basin (LIB). It is essential to monitor and regulate this pumping to sustain fresh groundwater lenses. In this study, we applied a modelling approach in combination with geostatistical analysis to identify the critical locations to monitor the groundwater levels for sustaining fresh groundwater in the LIB. Our approach included four steps: (i) simulating temporal heads using a calibrated hydrogeological model; (ii) sampling monitoring locations using a hexagonal pattern of sampling; (iii) applying principal component analysis (PCA) of the temporal head observations, and selecting high scoring locations from the PCA; and (iv) minimizing the observation points to represent the water level contours. The calibrated model was able to replicate the hydro-dynamic behavior of the study area, with a root mean square of 0.95 and an absolute residual mean of 0.74 m. The hexagonal pattern of spatial sampling resulted in a 195 point network, but PCA reduced this network to 135 points and contour classification reduced it even further to 59 points. The 195, 135, and 59 point networks represented the water levels with average standard errors of 0.098, 0.318, and 0.610 m, respectively. Long-term simulations with increased pumping showed that the water levels would best be assessed by 195 monitoring points, although 135 and 59 points would represent the depleting area but would not capture the water logging area. Full article

The geothermal characteristics of specific areas on Earth can be identified using geophysical and lithological logs based on deep boreholes, such as those more than 1000 m in depth. Based on the combined analyses of geophysical logs (temperature, caliper, electrical conductivity (EC), and natural gamma-ray logs) and lithological logs, as well as 1D steady-state heat transfer model, the deep groundwater flow and temperature were characterized in four deep boreholes (BH-1–BH-4) roughly 2000 m deep in the area of Heunghae-eup, Pohang, South Korea. The estimated thermal gradients from the temperature profiles are as follows: 22.37–30.77 °C/km for BH-1, 35.67–64.52 °C/km for BH-2, 40.85–46.44 °C/km for BH-3, and 33.33–35.71 °C/km for BH-4. According to the geophysical logs and lithology profiles, the groundwater mainly flows into and out of the boreholes through the basic dyke, rhyolite, and sandstone/mudstone. Evidently, the groundwater flows moving through the fractures and faults induce nonlinear temperature changes. The upward and downward groundwater flows passing through fractures and faults can be estimated using a 1D steady-state heat-transfer equation, by considering a fracture angle based on the lithological and geophysical profiles. To determine the direction (up/down) and rate of groundwater flow, the values for the parameter β were estimated as follows: −1.95 to 5.40 for BH-1, −13.48 to 4.87 for BH-2, −1.76 for BH-3, and −3.39 to 14.15 for BH-4. Full article

The effects of the unpredictability of the flow system in a karst area, with respect to bacterial communities in the aquatic environment, were evaluated. Bacterial communities from two different types of flow conditions (spring and sinkhole) were characterized and compared in the karst area. Proteobacteria, Bacteroidetes, and Verrucomicrobia were the major phyla present in the sampled spring and sinkhole waters. The water samples from the spring points were heavily loaded with Actinobacteria, especially Firmicutes, which accounted for 4.3% of the bacterial content. Furthermore, Actinobacteria were prevalent in some water samples, serving as indicators of the effects of seawater inundation. The richness and diversity of bacterial species were evaluated at the spring and sinkhole water points. The spring waters showed a higher bacterial richness and diversity compared to the sinkhole waters. Our results provide valuable information for the evaluation and investigation of microbial compositions in karst areas, which are characterized by heterogeneous hydrological conditions. The microbial species in karst areas are already exposed to changing hydrogeological conditions, and are likely to be confronted with future changes; thus, their spatiotemporal variations reflect the shifting baselines of physicochemical and ecosystem processes. Full article

In this paper, we studied the geo-hydrological structure and behavior of a reference catchment, located in the Cilento UNESCO Global Geopark, southern Italy, representative of the hilly, terrigenous and forested headwaters of the Mediterranean eco-region. Based on detailed hydrogeological and hydro-geomorphological surveys and geomorphometric analysis, starting in 2012, a hydro-chemical monitoring activity at the catchment and sub-catchment scale started, and a hydro-chemical dataset was progressively recorded at daily and sub-hourly time steps. Based on this dataset, the authors performed an original procedure to identify different runoff components, derived by applying cascade mass balance filtering. The integration of hydrological and geomorphological approaches allowed us to obtain an interesting conceptualization of the storm flow generation using hydro-chemical signatures related to different runoff components produced during the increasing–decreasing cycle of the flood event magnitude. The hydro-system activated progressively different runoff sources (i.e., groundwater, riparian corridor, hillslope and hollow) and involved various mechanisms (i.e., groundwater ridging, saturation-excess, infiltration-excess and soil pipe exfiltration). The geo-hydrological conceptualization was validated using a hysteresis Q-EC loop analysis performed on selected events that showed how hysteretic indices could be used to characterize the events in respect to their origins, mechanisms and pathways in similar catchments. Full article

The hydraulic and hydrogeological features of the Caposele aquifer have been investigated by using a numerical groundwater flow model. In particular, groundwater flow simulations were performed for a multilayered, unconfined aquifer in steady-state conditions for different thicknesses of the aquifer’s saturated zone. The Caposele groundwater model was carried out starting from a generic model drained by a unique spring outlet in accordance with the geo-hydrological features of the study area. The conceptual model was built considering hydrogeological features of spring catchment, and was then implemented with the MODFLOW numerical code. A combined 2D-3D approach was adopted, and the model was calibrated on borehole data available for the time period 2012–2019. Different thicknesses of the aquifer were set, and a reliable relationship was found between the hydraulic head, saturated zone and hydraulic conductivity of the aquifer. Using the MODPATH package, the mean travel time (Darcian) of groundwater was computed for five different scenarios, corresponding to the model’s depths; the analysis that was performed shows that the travel time is higher for a greater and lower for a smaller thickness of the aquifer’s saturated zone, respectively. The Caposele aquifer model was zoned in different sectors, named flow pipe areas, that play different roles in groundwater recharge-discharge processes. A vector analysis was also carried out in order to highlight the ascendant flow near the spring zone. Full article

Karst aquifer recharge areas are usually difficult to identify because of the complexity of these aquifers’ characteristics. On the other hand, their identification is very important in the aim of protecting the groundwater resources that these aquifers host. Regarding this topic, this paper presents an approach aimed at identifying karst aquifer recharge areas by the application of oxygen-18 and deuterium isotopes composition of groundwater coupled with hydrological features. Oxygen-18 and deuterium isotope composition of Capodacqua di Spigno Spring, in the South of the Latium Region, has been applied with rainfall and discharge values related to the feeding aquifer of this spring. As δ¹⁸O and δ²H values of groundwater samples are natural tracers of the recharge area’s elevation, we propose a model, based on the distribution of the basin surfaces involved as recharge areas, in relation to elevations. The model estimates, for any discharge value, the percentage of the topographic area involved in the aquifer recharge. The setting up of this simulated distribution is supported by a Weibull cumulative probability function. The results show that the measured discharges increase as larger areas with lower elevations are involved in the recharge process. Full article

Lowland river basins are characterised by complex hydrologic and hydraulic interactions between the different subsystems (aerated zone, groundwater, surface water), which may require physically-based dynamically-coupled surface water and groundwater hydrological models to reliably describe these processes. Exemplarily, for a typical north-eastern Germany lowland catchment (Tollense river with about 400 km²), an integrated hydrological model, MIKE SHE, coupled with a hydrodynamic model, MIKE 11, was developed and assessed. Hydrological and hydraulic processes were simulated from 2010 to 2018, covering strongly varying meteorological conditions. To achieve a highly reliable model, the calibration was performed in parallel for groundwater levels and river flows at the available monitoring sites in the defined catchment. Based on sensitivity analysis, saturated hydraulic conductivity, leakage coefficients, Manning’s roughness, and boundary conditions (BCs) were used as main calibration parameters. Despite the extreme soil heterogeneity of the glacial terrain, the model performance was quite reasonable in the different sub-catchments with an error of less than 2% for water balance estimation. The resulted water balance showed a strong dependency on land use intensity and meteorological conditions. During relatively dry hydrological years, actual evapotranspiration (ETa) becomes the main water loss component, with an average of 60%–65% of total precipitation and decreases to 55%–60% during comparatively wet hydrological years during the simulation period. Base flow via subsurface and drainage flow accounts for an approximate average of 30%–35% during wet years and rises up to 35%–45% of the total water budget during the dry hydrological years. This means, groundwater is in lowland river systems the decisive compensator of varying meteorological conditions. The coupled hydrologic and hydraulic model is valuable for detailed water balance estimation and seasonal dynamics of groundwater levels and surface water discharges, and, due to its physical foundation, can be extrapolated to analyse meteorological and land use scenarios. Future work will focus on coupling with nutrient transport and river water quality models. Full article

In this article, we present a meshless method based on the method of fundamental solutions (MFS) capable of solving free surface flow in three dimensions. Since the basis function of the MFS satisfies the governing equation, the advantage of the MFS is that only the problem boundary needs to be placed in the collocation points. For solving the three-dimensional free surface with nonlinear boundary conditions, the relaxation method in conjunction with the MFS is used, in which the three-dimensional free surface is iterated as a movable boundary until the nonlinear boundary conditions are satisfied. The proposed method is verified and application examples are conducted. Comparing results with those from other methods shows that the method is robust and provides high accuracy and reliability. The effectiveness and ease of use for solving nonlinear free surface flows in three dimensions are also revealed. Full article