HYDROLOGIC CLASSIFICATION OF STREAMS IN THE SOUTHEASTERN UNITED STATES

Hydrologic classification contributes to understanding the physical hydrology of a region by generalizing the complex and temporally variable streamflow sequences of different streams into groups that have shared characteristics and differences from each other. Stream classes can be useful in management applications when they reflect key functional characteristics such as streamflow generation processes, ecological structure, or sensitivity to human impacts. We developed two classification systems for rivers and streams in the southeastern United States based on 744 streamflow gaging sites with 20-year (minimum) streamflow records and relatively low likelihood of hydrologic alteration. The first classification represents seasonal streamflow regimes, while the second represents streamflow variability over multiple spatial and temporal scales.

Seasonal streamflow regimes were determined by clustering of mean monthly streamflow standardized by mean streamflow, which demonstrated six general types of streams that correspond to relatively distinct geographic regions. Multiple clustering approaches were applied to a set of multi-scale streamflow characteristics to identify key metrics of streamflow variability that were used to develop a rule-based classification. Purely statistical clustering uses the same set of flow metrics to define the flow characteristics for all classes, and hence tends to yield classes that have mixed characteristics. In contrast, rule-based classification enforced sharp physically-based distinctions in stream classes because it incorporated different combinations of flow metrics representing key hydrologic characteristics for each class. The multi-scale classification yielded stream classes that were more geographically dispersed than the seasonal classes.

Our results show that seasonal streamflow regimes, multi-scale streamflow variability, and streamflow magnitude are weakly coupled across the southeastern US, though each has management significance. As a result, multiple classification systems rather than a single, unified system may be needed to support a range of applications.