The time series analysis approach has the ability to recognize the characteristics of a karst aquifer by analyzing the response of the discharge from rainfall in a karst area. The karst aquifer is ...considered to be a system that influences the response to the conduit, fissure, and diffuse flow. This research was conducted in Mudal, Kiskendo, and Anjani Springs located in the Jonggrangan Karst, Java Island, Indonesia. The objective of this study was to determine the differences in the response of karst aquifers in releasing the flow in these three springs. The data used in this study are paired data of discharge and rainfall, which were obtained from January to November 2018 (
N
data = 31,408). Discharge data were obtained by installing an automatic water level logger, while rainfall data were obtained by installing automatic rain gauges in each spring. The time series analysis methods that were used are cross-correlation, auto-correlation, cross-amplitude, spectral density, phase functions, and gain functions. The master recession curve (MRC) calculation was performed to confirm the results from the time series analysis. The study reveals that each spring has a different response in releasing the flow components. Anjani Spring has the fastest response to conduit and fissure flows. However, this spring has the longest release duration of the diffuse flow component, indicating that a well-developed karst aquifer can also have high system memory and large storage capacity. Kiskendo Spring has a faster conduit and diffuse flow response than Mudal Spring but has the slowest fissure flow response, reflecting the void fissure that dominates in this site. In addition, MRC confirmed the results of the time series analysis calculation. This MRC analysis also shows that the study sites were categorized as a high sensitivity degree of karst aquifer. The karst aquifer’s characterization in this study can be used as basic data in the water resources management in the Jonggrangan Karst.
Understanding karst spring flow is important to accommodate the increasing water demand caused by the population growth and manage the freshwater water resource effectively. However, due to the ...spatial and temporal heterogeneity and complex hydrological processes in karst systems, predicting karst spring discharge remains challenging. In this study, three deep learning‐based models, including long short‐term memory (LSTM), gated recurrent unit (GRU) and simple recurrent neural network (RNN), are framed with an encoder–decoder architecture to provide multiple‐step‐ahead spring discharge prediction. The encoder–decoder architecture includes an encoder that reads and encodes the input sequence into a vector and decoder that deciphers the vector and outputs the predicted sequence. Three hybrid models called LSTM‐ED, GRU‐ED and simple RNN‐ED are compared with single‐step models and multiple‐step models without the encoder–decoder architecture to investigate the role of the encoder–decoder architecture on multi‐step‐ahead prediction. The sensitivity of the selection of input time and lead time steps on the karst spring discharge prediction is evaluated. The predicted results are compared with the observed spring discharge. It implies that: (1) LSTM‐ED, GRU‐ED and RNN‐ED models obtain similar results on predicting karst spring discharge multiple time steps ahead; (2) three hybrid multiple‐step models outperform the single‐step models in making consistent and accurate spring discharge predictions; (3) the multiple‐step models framed with an encoder–decoder architecture obtain better spring discharge prediction results than the single‐step models and multiple‐step models without the encoder–decoder structure; (4) the LSTM‐ED, GRU‐ED and simple RNN‐ED models are sensitive to the selection of lead time and insensitive to the selection of input time step. A short lead time typically yields a more accurate spring discharge prediction.
The artificial neural networks (ANN)‐based model with the encoder–decoder architecture: m time step input and m time step output.
Changbai Mountain area is an important mineral water storage and development area in China. The hydrochemical composition of mineral water is the decisive factor for mineral–water quality. Based on ...the hydrochemical data of 74 mineral water samples collected from 2018 to 2020, the characteristics and formation mechanism of the hydrochemical components of the mineral water were analyzed. The results show that the formation of single-type mineral springs (metasilicate mineral water) is controlled by rock weathering; compound mineral springs (metasilicate mineral water with high CO
2
content) are the product of CO
2
-rich, weakly acidic, confined hot groundwater with high salinity, which are mixed with shallow groundwater as rising along the fracture. The volcanic geological process greatly influences the formation of the hydrochemical components of mineral springs on the North slope of Changbai Mountain. The mineral springs on the Longgang Mountain are greatly affected by human activities. The results of cluster analysis only consider that hydrochemical components are consistent with the classification of the areas which concentrated distributions of mineral as determined by hydrogeological and geomorphological studies. The results of this study are useful for understanding the distribution, hydrochemical characteristics, and formation mechanism of mineral springs in the Changbai Mountain area of China and provide the theoretical basis for the protection and development of mineral spring.
Thermal springs of Aravali, Tural and Rajawadi are located in the Deccan volcanic province (DVP) in the western part of Maharashtra, India and are covered by Deccan basalts. These springs run ...parallel to the Western Ghats and geochemical studies denote that these hot water springs are of meteoritic origin that emerge from basement rock.
To understand the geoelectric structure and possible source zone of hot water springs, a 26-station audiomagnetotelluric (AMT) survey was carried out along E W profiles across Aravali, Tural and Rajawadi geothermal springs with a station spacing of about 1–1.5 km. A 2D inversion was carried out jointly for transverse electric (TE) and transverse magnetic (TM) data after the distortion and decomposition analysis.
These geothermal zones appear as high conductive zones at shallow depth and are associated with fault/fracture zones within the sedimentary basin. The basement depth increases from 1 km to 2.5 km moving from Aravali to Rajawadi geothermal zones. Volcanic plugs at a shallow depth beneath Tural and Rajawadi thermal springs act as a source rock for heat. Thus, hot water temperatures are higher for these two thermal springs relative to Aravali thermal spring, which is devoid of magma intrusion and related to circulation of meteoric water over basement.
•AMT studies across geothermal zones in western Maharashtra, India.•Conductivity zones (Fault/fracture) signifies geothermal fluid channel.•Volcanic plug acts as a source rock for heat in Tural and Rajawadi springs.•Hot springs are linked to flow of meteoric water through the basement/source rock.
The archival records of chemical composition of mineral waters in Szczawno-Zdrój spa were analyzed in terms of variation of ionic ratios to explain the possible source and origin of the major ...compounds dissolved in water and evolution of groundwater chemical composition in time. The analyzed data contained the longest available series of chemical records, dating back to 1962, and related to waters discharged by five main springs: Dąbrówka, Marta, Młynarz, Mieszko and Mieszko 14. The research showed that mineral waters in Szczawno-Zdrój belong to shallow meteoric CO
2
-rich, Rn-containing groundwaters which form their chemical composition mainly through the interaction with aquifer rocks. Detailed analysis of long-term variation of ionic ratios revealed that (1) the carbonates weathering, mostly acid hydrolysis of limestones and dolomites, and (2) the ion exchange reactions with clay minerals, mainly the so-called natural softening, play a fundamental role in formation of the chemical composition of studied waters. Both processes are responsible for the occurrence of dominant ions in solution such as Ca
2+
, Mg
2+
, Na
+
, and HCO
3
−
. The aluminosilicates hydrolysis occurs with variable extent, but plays rather secondary role in formation of chemical composition. The time distributions of major element concentrations in studied waters showed a characteristic “concave” shape, indicating the decrease in concentrations beginning in the 60s and ending around 2005–2010. Such “concave” shape trends are not reflected in time distribution of ionic ratios which strongly suggests the occurrence of a simple dilution of chemical composition of mineral waters by the influx of fresh water. The observed considerable fluctuations of chemical composition of mineral waters in Szczawno-Zdrój are most probably associated with climatic factors, namely: the increased amounts of atmospheric precipitation in particular periods of time and its seasonal distribution. Such influx of fresh waters reduces considerably mineralization of shallow groundwaters and directly increases springs discharge.
The Dead Sea is a closed lake, the water level of which is lowering at an alarming rate of about 1 m/year. Factors difficult to determine in its water balance are evaporation and groundwater inflow, ...some of which emanate as submarine groundwater discharge. A vertical buoyant jet generated by the difference in densities between the groundwater and the Dead Sea brine forms at submarine spring outlets. To characterize this flow field and to determine its volumetric discharge, a system was developed to measure the velocity and density of the ascending submarine groundwater across the center of the stream along several horizontal sections and equidistant depths while divers sampled the spring. This was also undertaken on an artificial submarine spring with a known discharge to determine the quality of the measurements and the accuracy of the method. The underwater widening of the flow is linear and independent of the volumetric spring discharge. The temperature of the Dead Sea brine at lower layers primarily determines the temperature of the surface of the upwelling, produced above the jet flow, as the origin of the main mass of water in the submarine jet flow is Dead Sea brine. Based on the measurements, a model is presented to evaluate the distribution of velocity and solute density in the flow field of an emanating buoyant jet. This model allows the calculation of the volumetric submarine discharge, merely requiring either the maximum flow velocity or the minimal density at a given depth.
A vertical buoyant jet generated by the difference in densities between the groundwater and the Dead Sea brine forms at submarine spring outlets. The velocity and density of the ascending submarine groundwater across the center of the stream along several horizontal sections and equidistant depths were measured while divers sampled the spring. A model was developed to allow the calculation of the volumetric submarine discharge, merely requiring either the maximum flow velocity or the minimal density at a given depth.
Karst springs are essential drinking water resources, however, modeling them poses challenges due to complex subsurface flow processes. Deep learning models can capture complex relationships due to ...their ability to learn non‐linear patterns. This study evaluates the performance of the Transformer in forecasting spring discharges for up to 4 days. We compare it to the Long Short‐Term Memory (LSTM) Neural Network and a common baseline model on a well‐studied Austrian karst spring (LKAS2) with an extensive hourly database. We evaluated the models for two further karst springs with diverse discharge characteristics for comparing the performances based on four metrics. In the discharge‐based scenario, the Transformer performed significantly better than the LSTM for the spring with the longest response times (9% mean difference across metrics), while it performed poorer for the spring with the shortest response time (4% difference). Moreover, the Transformer better predicted the shape of the discharge during snowmelt. Both models performed well across all lead times and springs with 0.64–0.92 for the Nash–Sutcliffe efficiency and 10.8%–28.7% for the symmetric mean absolute percentage error for the LKAS2 spring. The temporal information, rainfall and electrical conductivity were the controlling input variables for the non‐discharge based scenario. The uncertainty analysis revealed that the prediction intervals are smallest in winter and autumn and highest during snowmelt. Our results thus suggest that the Transformer is a promising model to support the drinking water ion management, and can have advantages due to its attention mechanism particularly for longer response times.
Key Points
The Transformer architecture was applied in karst hydrology for the first time, showing high performance for discharge forecasting
Monte Carlo dropout revealed that the prediction intervals are smallest and cover the measured discharges best in winter and autumn
The high temporal resolution of the input data sets improved the forecasting performance
Knowledge of the hydraulic and geological properties of karst systems is particularly valuable to hydrogeologists because these systems represent an important source of potable water in many ...countries. However, the high heterogeneity that characterizes karst systems complicates the definition of karst hydrogeological properties, and their estimation involves complex and expensive techniques. In this study, a workflow for karst spring characterization was used to analyze two springs, Nanto spring and Mossano spring, located in the Berici Mountains (NE Italy). Based on the data derived from 4 years of continuous hourly monitoring of discharge, water temperature and specific electrical conductivity, a hydrogeological conceptual model for the monitored springs was proposed. Flow rate measurements, which combined recession curve, flow duration curve and autocorrelation function techniques, were used to evaluate the spring discharge variability. Changes in spring discharge can be related both to the degree of karstification/permeability and to the size of the karst aquifer. Moreover, combining monitored parameters and rainfall—analyzed by the cross-correlation function and VESPA (Vulnerability Estimator for Spring Protection Areas) index approach—permitted assessment of the spring response to recharge and the behavior of the drainage system. Although the responses to the recharge events were quite similar, the two springs showed some differences in terms of the degree of karstification. In fact, Mossano spring showed a more developed karst system than Nanto spring. Three systems (two karsts and one matrix/fractured) are outlined for Mossano spring, while two systems (one karst and one matrix/fractured) are outlined for Nanto spring.