Application of multivariate analysis to suspended matter particle size distribution in a karst aquifer

Numerous quantitative and qualitative variables control suspended sediment dynamics in karst systems. The objective of this study was to identify the hydrodynamic variables controlling the transport properties of particles in a karst aquifer (western Paris basin). The particle size distribution of suspended sediment infiltrating via a swallow hole was compared to that in discharge from a spring, allowing identification of the particle transport properties of the karst system. Hill and Smith analysis, a type of multivariate analysis that allows joint examination of quantitative and qualitative variables, was used to identify the hydrodynamic parameters controlling the transport properties of the suspended matter. The results demonstrate that the particle size distribution discharging at the karst spring is controlled by spring discharge and the hydraulic gradient of the system. The hydraulic gradient is defined by the piezometric level and the stage of the Seine River, which is in turn controlled by the tide. This study illustrates the use of Hill and Smith analysis to identify those variables which control suspended sediment transport. It also illustrates the application of this analysis to identify boundary conditions and evaluate variables which control the behaviour of the hydrologic system. Copyright © 2007 John Wiley & Sons, Ltd.     

Investigation of Transport Processes inside Karst Aquifer by Means of STATIS

Environmental data sets are often multidimensional and consequently display complex structure. This article shows the limitations of principal component analysis (PCA) for the study of such three-dimensional (3D) data sets. These limitations can be resolved by the use of the statistical tool STATIS. The inlet (a swallow hole) and the outlet (a spring) of a karst system of the Western Paris basin were sampled during three rain events of various intensities. These 3D geochemical data sets (variables × sites × dates) for a karst system were analyzed by STATIS method to identify hydrological processes. STATIS proceeds in three steps (interstructure, compromise, and intrastructure), which allows us to focus the analysis of hydrologic systems at different temporal and spatial scales. Compromise plane shows that suspended matter and flood are not simultaneous and highlights a rapid flow, characterized by turbidity and phosphate, which represents a point source contamination, and a ground water flow contaminated by nitrate. Intrastructure plane allows us to compare hydrochemical variations between the swallow hole and the spring lead. By this way, hydrological processes such as direct transfer and resuspension of intrakarstic sediments before and after the flood were identified what cannot be realized by comparison of inlet and outlet breakthrough curves. Finally, results obtained from the same data set by STATIS and a coupled study using PCA and normalized hysteresis curves were compared. This comparison shows the efficiency of STATIS at the identification of transport processes and vulnerability of karst system and its potential for hydrological applications.     

Anomalous behaviour of specific electrical conductivity at a karst spring induced by variable catchment boundaries: the case of the Podstenjšek spring,Slovenia

Anomalous behaviour of specific electrical conductivity (SEC) was observed at a karst spring in Slovenia during 26 high‐flow events in an 18‐month monitoring period. A conceptual model explaining this anomalous SEC variability is presented and reproduced by numerical modelling, and the practical relevance for source protection zoning is discussed. After storm rainfall, discharge increases rapidly, which is typical for karst springs. SEC displays a first maximum during the rising limb of the spring hydrograph, followed by a minimum indicating the arrival of freshly infiltrated water, often confirmed by increased levels of total organic carbon (TOC). The anomalous behaviour starts after this SEC minimum, when SEC rises again and remains elevated during the entire high‐flow period, typically 20–40 µS/cm above the baseflow value. This is explained by variable catchment boundaries: When the water level in the aquifer rises, the catchment expands, incorporating zones of groundwater with higher SEC, caused by higher unsaturated zone thickness and subtle lithologic changes. This conceptual model has been checked by numerical investigations. A generalized finite‐difference model including high‐conductivity cells representing the conduit network (“discrete‐continuum approach”) was set up to simulate the observed behaviour of the karst system. The model reproduces the shifting groundwater divide and the nearly simultaneous increase of discharge and SEC during high‐flow periods. The observed behaviour is relevant for groundwater source protection zoning, which requires reliable delineation of catchment areas. Anomalous behaviour of SEC can point to variable catchment boundaries that can be checked by tracer tests during different hydrologic conditions. Copyright © 2011 John Wiley & Sons, Ltd.     

Effect of Bank Deformations on Pollutant Transport in Rivers in Cryolithozone: Laboratory and Mathematical Modeling

The study is focused on pollution transport in rivers flowing in cryolithozone and takes into account possible channel deformations caused by thawing of permafrost rocks in which the channel runs. The methods used are laboratory and mathematical modeling. The mathematical model consists of four blocks: hydrodynamic, thermal, deformation, and pollutant transport. The model was validated and tested against the data of a laboratory experiment. Numerical experiments suggested conclusions regarding the propagation of pollutants entering the flow from sources in thawing bank slopes and on river floodplain.     

Comparison of flowpaths to a well and spring in a karst aquifer

The permeability of some karst aquifers consists of networks of poorly integrated conduits and dissolution-widened fractures. The flow includes conduit flow, especially during storm recharge, but lacks the focused recharge into single master conduits that occurs in more highly developed karst systems. The proportions of conduit and dispersed flow are difficult to quantify in such systems. This study examines the flowpaths in a small karst watershed, based on comparing the physical and chemical response to storm flow at both a spring and a well. By conducting continuous monitoring at both locations, a better understanding of the flowpaths in a poorly integrated network was obtained. A more permeable flowpath to the spring leads to faster storm response and lower ion concentrations. The flowpath to and from the well is more complicated. The higher ion content and slower storm response suggest slower, more dispersed flowpaths. However, the well has greater variation in ion chemistry. Periodic recharge may dilute well concentrations due to faster (conduit or fracture) flowpaths. Although karst systems such as this are difficult to characterize, applying a variety of geochemical and physical monitoring techniques at multiple locations illustrates that the flowpaths can vary in both space and time.     

Assessing the contamination of water flow in Soldatskaya Cave (Crimea)

The contamination of water flow in Soldatskaya Cave is studied. Groundwater of the cave was found to feature bacterial pollution. The distribution of bacterial cells over the karst system is examined. The contamination is shown to correlate with anthropogenic load. The major hydrodynamic and physicochemical characteristics of groundwater flow are evaluated, and their correlation with self-purification is demonstrated. Components of the karst water self-purification mechanisms are analyzed. The possibility to intensify karst water self-purification is considered.     

Transport of suspended solids from a karstic to an alluvial aquifer: the role of the karst/alluvium interface

This study focuses on the coupled transport of dissolved constituents and particulates, from their infiltration on a karst plateau to their discharge from a karst spring and their arrival at a well in an alluvial plain. Particulate markers were identified and the transport of solids was characterised in situ in porous and karstic media, based on particle size analyses, SEM, and traces. Transport from the sinkhole to the spring appeared to be dominated by flow through karst: particulate transport was apparently conservative between the two sites, and there was little difference in the overall character of the particle size distribution of the particulates infiltrating the sinkhole and of those discharging from the spring. Qualitatively, the mineralogy of the infiltrating and discharging material was similar, although at the spring an autochthonous contribution from the aquifer was noted (chalk particles eroded from the parent rock by weathering). In contrast, transport between the spring and the well appears to be affected by the overlying alluvium: particles in the water from the well, showed evidence of considerable size-sorting. Additionally, SEM images of the well samples showed the presence of particles originating from the overlying alluvial system; these particles were not found in samples from the sinkhole or the spring. The differences between the particulates discharging from the spring and the well indicate that the water pumped from the alluvial plain is coming from the karst aquifer via the very transmissive, complex geologic interface between the underlying chalk formation and the gravel at the base of the overlying alluvial system.     

Determination of the conduit structure in a karst aquifer based on tracer data—Lurbach system,Austria

A structure model was used to analyse solute‐transport parameter estimates based on tracer breakthrough curves. In the model system, groundwater flow is envisioned to be organised in a complex conduit network providing a variety of short circuits with relative small carrying capacities along different erosion levels. The discharge through the fully filled conduits is limited owing to void geometries and turbulent flow; thus, a hierarchic overflow system evolves where conduits are (re‐)activated or dried up depending on the flow condition. Exemplified on the Lurbach–Tanneben karst aquifer, the applicability of the model approach was tested. Information derived from multi‐tracer experiments performed at different volumetric flow rates enabled to develop a structural model of the karst network, under constraint of the geomorphological and hydrological evolution of the site. Depending on the flow rate, groundwater is divided into up to eight flow paths. The spatial hierarchy of flow paths controls the sequence of flow path activation. Conduits of the topmost level are strongly influenced by reversible alteration processes. Sedimentation or blocking causes an overflow of water to the next higher conduit. Flow path specific dissolutional denudation rates were estimated using the temporal development of the partial discharge. Copyright © 2012 John Wiley & Sons, Ltd.     

Can Karst Conduit Models Be Calibrated? A Dual Approach Using Dye Tracing and Temperature

Modeling flow and transport using both temperature and dye tracing provides constraints that can improve understanding of karst networks. A laminar flow and transport model using the finite element subsurface flow model simulated the conduit connection between a sinking stream and spring in central Pennsylvania to evaluate how conduit morphology might affect dye transport. Single and overly tortuous conduit models resulted in high concentrations as dye flowed back into the conduit from the matrix after dye injections ceased. A forked conduit model diverted flow from the main conduit, reducing falling limb dye concentration. Latin hypercube sampling was performed to evaluate the sensitivity of 52 parameter combinations (conduit hydraulic conductivity, conduit cross-sectional area, matrix transmissivity, matrix porosity, and dispersivity) for four conduit geometry scenarios. Sensitivity of arrival time for 50% of the dye indicated no parameter combinations which simulate falling limb dye concentrations for tortuous geometries, confirming the importance of the forked geometry regardless of other parameters. Temperature data from high-resolution loggers were then incorporated into the forked conduit model to reproduce seasonal spring temperature using variable sink inflow. Unlike the dye trace models, the thermal models were sensitive to other model parameters, such as conduit cross-sectional area and matrix transmissivity. These results showed this dual approach (dye and temperature) to karst network modeling is useful for (1) exploring the role of conduit and matrix interaction for contaminant storage, (2) constraining karst conduit geometries, which are often poorly understood, and (3) quantifying the effect of seasonal trends on karst aquifers.     

Simulation of spring flows from a karst aquifer with an artificial neural network

In China, 9·5% of the landmass is karst terrain and of that 47,000 km² is located in semiarid regions. In these regions the karst aquifers feed many large karst springs within basins of thousands of square kilometres. Spring discharges reflect the fluctuation of ground water level and variability of ground water storage in the basins. However, karst aquifers are highly heterogeneous and monitoring data are sparse in these regions. Therefore, for sustainable utilization and conservation of karst ground water it is necessary to simulate the spring flows to acquire better understanding of karst hydrological processes. The purpose of this study is to develop a parsimonious model that accurately simulates spring discharges using an artificial neural network (ANN) model. The karst spring aquifer was treated as a non‐linear input/output system to simulate the response of karst spring flow to precipitation and applied the model to the Niangziguan Springs, located in the east of Shanxi Province, China and a representative of karst springs in a semiarid area. Moreover, the ANN model was compared with a previous time‐lag linear model and it was found that the ANN model performed better. Copyright © 2007 John Wiley & Sons, Ltd.     

Influence of conduit network geometry on solute transport in karst aquifers with a permeable matrix

In karst aquifers with significant matrix permeability, water and solutes are exchanged between the conduits and carbonate matrix. Transport through the matrix increases the spread of solutes and increases travel times. This study numerically evaluates advective solute transport in synthetic karst systems that contain 3D branching conduit networks. Particle tracking is performed to analyze the spatial and temporal transport history of solute that arrives at the conduit outlet. Three measures of transport connectivity are used to quantify the solute migration behavior: the skewness of the particle arrival time distribution, the normalized fifth percentile of arrival times, and the fraction of the total travel time that occurs within conduits. All three of these metrics capture the influence of conduit network geometry on solute transport. A more tortuous network leads to enhanced conduit-matrix mixing, which reduces the transport connectivity and yields a broader distribution of solute arrival times. These results demonstrate that the conduit network geometry is an important control on solute transport in karst systems with a permeable matrix.     

Vertical distribution of a pollutant in river flow: Mathematical modeling

A mathematical model has been developed to assess the role of the shape of channel cross-section, the position of pollutant source on the bank slope, and the sedimentation rate of pollutant particles in the process of pollutant transport in water flow. The procedure of numerical experiments and the obtained results and conclusions can be of use in the simulations and studies of pollution propagation from sources on riverbanks in permafrost zone, which are subject to thermal erosion.     

Comparison of artificial neural network models for hydrologic predictions at multiple gauging stations in an agricultural watershed

This paper reports on an evaluation of the use of artificial neural network (ANN) models to forecast daily flows at multiple gauging stations in Eucha Watershed, an agricultural watershed located in north‐west Arkansas and north‐east Oklahoma. Two different neural network models, the multilayer perceptron (MLP) and the radial basis neural network (RBFNN), were developed and their abilities to predict stream flow at four gauging stations were compared. Different scenarios using various combinations of data sets such as rainfall and stream flow at various lags were developed and compared for their ability to make flow predictions at four gauging stations. The input vector selection for both models involved quantification of the statistical properties such as cross‐, auto‐ and partial autocorrelation of the data series that best represented the hydrologic response of the watershed. Measured data with 739 patterns of input–output vector were divided into two sets: 492 patterns for training, and the remaining 247 patterns for testing. The best performance based on the RMSE, R² and CE was achieved by the MLP model with current and antecedent precipitation and antecedent flow as model inputs. The MLP model testing resulted in R² values of 0·86, 0·86, 0·81, and 0·79 at the four gauging stations. Similarly, the testing R² values for the RBFNN model were 0·60, 0·57, 0·58, and 0·56 for the four gauging stations. Both models performed satisfactorily for flow predictions at multiple gauging stations, however, the MLP model outperformed the RBFNN model. The training time was in the range 1–2 min for MLP, and 5–10 s for RBFNN on a Pentium IV processor running at 2·8 GHz with 1 MB of RAM. The difference in model training time occurred because of the clustering methods used in the RBFNN model. The RBFNN uses a fuzzy min‐max network to perform the clustering to construct the neural network which takes considerably less time than the MLP model. Results show that ANN models are useful tools for forecasting the hydrologic response at multiple points of interest in agricultural watersheds. Copyright © 2008 John Wiley & Sons, Ltd.     

Transport and Variability of Fecal Bacteria in Carbonate Conglomerate Aquifers

Clastic sedimentary rocks are generally considered non‐karstifiable and thus less vulnerable to pathogen contamination than karst aquifers. However, dissolution phenomena have been observed in clastic carbonate conglomerates of the Subalpine Molasse zone of the northern Alps and other regions of Europe, indicating karstification and high vulnerability, which is currently not considered for source protection zoning. Therefore, a research program was established at the Hochgrat site (Austria/Germany), as a demonstration that karst‐like characteristics, flow behavior, and high vulnerability to microbial contamination are possible in this type of aquifer. The study included geomorphologic mapping, comparative multi‐tracer tests with fluorescent dyes and bacteria‐sized fluorescent microspheres, and analyses of fecal indicator bacteria (FIB) in spring waters during different seasons. Results demonstrate that (1) flow velocities in carbonate conglomerates are similar as in typical karst aquifers, often exceeding 100 m/h; (2) microbial contaminants are rapidly transported toward springs; and (3) the magnitude and seasonal pattern of FIB variability depends on the land use in the spring catchment and its altitude. Different groundwater protection strategies that currently applied are consequently required in regions formed by karstified carbonatic clastic rocks, taking into account their high degree of heterogeneity and vulnerability.     

Using Stormwater Hysteresis to Characterize Karst Spring Discharge

Discharge from karst springs contains a mixture of conduit and matrix water, but the variations in groundwater mixing are poorly known. Storm events present an opportunity to try to map flow components because water entering during storms is more dilute and provides a tracer as it mixes with pre‐event water along the flowpath from the recharge area to discharge at a spring. We used hysteresis plots of Mg/Ca ratios in a spring in the Cumberland Valley of Pennsylvania to map conduit (higher Ca) vs. diffuse (higher Mg) sources of recharge. We observed two types of temporal heterogeneity: within a storm event and from storm to storm. The timing of the variation in Mg/Ca suggested sources of mixing waters. An increase in the Mg/Ca ratio at the beginning of some storms while conductivity declined suggested diffuse recharge through the epikarst. The rapid changes in Mg/Ca ratios for low‐intensity events probably occurred as the rainfall waxed and waned and illustrate that a variety of flowpaths are available at this spring because additional flushing of Mg occurred. In contrast, the conductivity hysteresis began with dilute water initially and rotation was similar from storm to storm. Hysteresis plots of the Mg/Ca ratio have the potential of revealing more of the complexity in discharge than conductivity alone. A better understanding of flow components in karst is needed to protect these aquifers as a groundwater resource.     

Analysis of karst aquifer spring flows with a gray system decomposition model

There are approximately 470,000 km(2) of karst aquifers that feed many large springs in North China. Turbulent flow often exists in these karst aquifers, which means that the classical ground water model based on Darcy's law cannot be applied here. Ground water data are rare for these aquifers. As a consequence, it is difficult to quantitatively investigate ground water flow in these karst systems. The purpose of this study is to develop a parsimonious model that predicts karst spring discharge using gray system theory. In this theory, a white color denotes a system that is completely characterized and a black color represents a system that is totally unknown. A gray system thus describes a complex system whose characteristics are only partially known or known with uncertainty. Using this theory, we investigated the karst spring discharge time series over different time scales. First, we identified three specific components of spring discharge: the long-term trend, periodic variation, and random fluctuation. We then used the gray system model to simulate the long-term trend and obtain periodic variation and random fluctuation components. Subsequently, we developed a predictive model for karst spring discharge. Application of the model to Liulin Springs, a representative example of karst springs in northern China, shows that the model performs well. The predicted results suggest that the Liulin Springs discharge will likely decrease over time, with small fluctuations.     

A global experimental system approach of karst springs' hydrographs and chemographs

Investigation techniques for karst flow systems are based mainly on the study of different signals leaving the system caused by natural or induced external influences. Each signal represents one of the systems outputs (e.g., hydraulic, chemical, physical, or isotopic responses) that reflect the characteristics of the entire system. In this paper, we present a method to infer information about the structure of karst systems. It is based on a simultaneous analysis of chemical and hydraulic responses. Beside the classical piston flow at the beginning of a flood pulse, we define a chemically based recession flow phase. During this phase, field data show that the concentration of total dissolved solids can be considered as an exponential function of the logarithm of flow. This relationship allows two parameters to be defined, one of which is dependent on the structure and degree of development of the karst conduit network, the other is dependent mainly on bioclimatic factors. Data collected from seven karst springs are used to support ideas introduced in the paper.     

Validation of a Numerical Indicator of Microbial Contamination for Karst Springs

Rapid changes in spring water quality in karst areas due to rapid recharge of bacterially contaminated water are a major concern for drinking water suppliers and users. The main objective of this study was to use field experiments with fecal indicators to verify the vulnerability of a karst spring to pathogens, as determined by using a numerical modeling approach. The groundwater modeling was based on linear storage models that can be used to simulate karst water flow. The vulnerability of the karst groundwater is estimated using such models to calculate criteria that influence the likelihood of spring water being affected by microbial contamination. Specifically, the temporal variation in the vulnerability, depending on rainfall events and overall recharge conditions, can be assessed and quantified using the dynamic vulnerability index (DVI). DVI corresponds to the ratio of conduit to diffuse flow contributions to spring discharge. To evaluate model performance with respect to predicted vulnerability, samples from a spring were analyzed for Escherichia coli, enterococci, Clostridium perfringens, and heterotrophic plate count bacteria during and after several rainfall events. DVI was shown to be an indication of the risk of fecal contamination of spring water with sufficient accuracy to be used in drinking water management. We conclude that numerical models are a useful tool for evaluating the vulnerability of karst systems to pathogens under varying recharge conditions     

Analytical and experimental study on dissolved pollutant wash‐off over impervious surfaces

Non‐point source pollution in the impervious surface of city, which including dissolved and particulate pollutants, is a significant source of water pollution. Simple first‐order decay models can generally simulate the cumulative wash‐off process of the particulate pollutants. There is inadequate knowledge as to whether or not they are suitable for dissolved pollutants. This study presents a mathematical wash‐off model for dissolved pollutants, which combines analytical equations for overland flows and the exponential equation for the pollutant wash‐off. A series of laboratory experiments have been conducted to verify this wash‐off model. It shows that the pollutant concentration and pollutant transport rate can be predicted well by the newly developed equations. It is found that the pollutant concentration monotonically decreases to zero as the accumulated pollutants are washed off, whereas the pollutant transport rate first increases to the maximum value and then decreases to zero. The maximum pollutant transport rate is found to increase with the decrease of the arrival time of the maximum value. The difference between the simplified exponential model and the amended wash‐off equation depends on the initial residual percentage (P_c), but the present equation generally provides a more accurate representation of the wash‐off process of dissolved pollutants.