Abstract:

This study was undertaken to gain a better understanding of karst hydrology. To do this, the present day hydrology and the paleohydrology were determined in three karst basins. The basins chosen were the Swago, Locust and Spring Creek basins in Pocahontas and Greenbrier Counties, West Virginia. A number of conventional field techniques were used successfully in this study, including the following: current meter and dye dilution gauging; dye and lycopodium stream tracing; geological and cave mapping; the setting up of stage recorders; geochemistry; and limestone erosion measurements. The climate of the region was investigated to obtain realistic precipitation, temperature and potential evaporation data over the study basins.
It was found that the mean precipitation over two of the basins was 30% higher than recorded data in the valleys. The karst development of the basins was found to take place in four major stages. These were: A) initial surficial flow, B) strike controlled drainage, C) major piracies from one sub-basin to another, and D) shortening of the flow routes. The major controls on the karst development were found to be: A) the Taggard shale, B) the strike direction, which controlled early basin development, and C) the hydraulic gradient from the sink to rising, which controlled later basin development.
To better assess the quantitative hydrology, and to assist in determining the type of unexplorable flow paths, a watershed model was developed. This modelled the streamflow from known climatic inputs using a number of measured or optimized parameters. The simulation model handled snowmelt, interception, infiltration, interflow, baseflow, overland flow, channel routing, and evaporation from the interception, soil water, ground water, snowpack and channel water. The modelled basin could be split up into 20 segments, each with different hydrological characteristics, but a maximum of 3 segments was used in this study.
A total of 29 parameters was used in the model although only 10 (other than those directly measurable) were found to be sensitive in the three basins. The simulated streamflow did not match the real flows very well due to errors in the data input and due to simplifications in the model. It was found, however, that as the proportion of the limestone in a segment increased the overland flow decreased, the interflow increased, the baseflow and interflow recessions were faster, the soil storages were smaller and the infiltration rate was higher, than in segments with a larger proportion of exposed clastics. The flow characteristics of the inaccessible conduits were inferred from the channel routing parameters and it was postulated that the majority of the underground flow in the karst basins was taking place under vadose conditions.