An alternative approach to investigation of sediment transport through a karst aquifer.

• Input and output time series from the conduit system of a karst aquifer are analyzed. • A simple water-balance model of epikarst zone is proposed. • Hourly internal runoff and diffuse drainage from epikarst zone are estimated. • Correlation and partial correlation functions are used. • A more precise insight into the sediment transport process in aquifer is obtained. Information about the transport of suspended sediment through a karst aquifer is obtained by the time series analysis of the input and outputs from the conduit system of vadose and phreatic zone. The input is the internal runoff from epikarst zone at the analyzed location. The outputs are the discharge from karst spring, as well as the turbidity and electrical conductivity of spring water. The diffuse drainage from epikarst zone is considered as a control process that changes the quantity of water in the conduit system, and consequently it affects the outputs. The time series of internal runoff and diffuse drainage are obtained by a simple water-balance model of epikarst zone based on the concept of saturation threshold. The proposed model has two parameters: maximal capacity of epikarst storage, and maximal diffuse drainage from epikarst storage, which are estimated by using the cumulative rainfall and turbidity data, as well as the value of runoff coefficient. The time series are analyzed by means of the correlation and partial correlation functions. The presented results show that the overall capability and accuracy of the time series analysis are significantly improved by considering the conduit system separately. It is possible to register and to characterize processes in karst aquifer that are not recognizable by using classical approaches. The results for Jadro Spring show that the memory of hypothetical discharge generated by the internal runoff is practically same as the memory of time series of turbidity. The memory of hypothetical discharge generated by the diffuse drainage is very similar to the memory of time series of discharge. The duration of turbidity response is 22 days, which entirely corresponds to the duration of fresh rainwater withdrawal from the conduit system. During the first 5 days, turbidity is generated mostly by the internal runoff from the central part of catchment. During the last 12 days, turbidity has been generated by the internal runoff from the eastern part of catchment. Two components in turbidity are recognized, which are manifested as two peaks in the turbidity graph. Both peaks are produced dominantly by autochthonous sediment. [ABSTRACT FROM AUTHOR]