Separating Anthropogenic Influences on Hydrology Using Data Mining Techniques – A Case Study of the Congaree National Park, South Carolina

The Congaree National Park is located along the Congaree River, 25 miles downstream from the confluence of the Saluda and Broad Rivers. The Park includes extensive swamps of old growth bottomland hardwood forest. Since 1930, the Congaree River has been regulated by the Saluda Dam on the Saluda River. Many ecologists and water-resource managers have hypothesized that the regulated flows on the Saluda River have substantially decreased the magnitude and frequency of flooding of the riparian wetlands in the Park.The effects of the Saluda Dam on the water levels in the Congaree River were determined using data-mining techniques including statistical analysis of peak flows and the development of empirical artificial neural network models to analyze 75 years of daily gage heights. Flood frequency analysis on the peak flows on the Broad, Saluda, and Congaree Rivers was done for various periods of the historic record for periods of pre- and post-impoundment. The peak flows for the regulated Congaree River in South Carolina were compared to the unregulated Broad River in Georgia to evaluate potential changes in peak flows due to climate variability. Artificial neural network models were developed using 3 years of pre-regulated daily mean flow data (1926 to 1929) to synthesize a 75-year (1930 to 2005) water-level hydrograph of a "no dam" or "no regulation" condition to compare with the actual regulated flows.The analysis of peak flows indicated that the differences in the magnitude and frequency of floods after the construction of the Saluda Dam are more a result of the major floods that occurred in the early 1900s rather than the operation of the dam. The analysis of daily gage heights indicated that the operation of the dam has decreased high gage heights that occur in the first 6 months of the year and increased the low gage heights that occur in the last half of the year. Operation of the dam has affected monthly average gage heights by up to 18 percent.