Spatial and temporal patterns in conterminous United States streamflow characteristics

Spatial and temporal patterns in annual and seasonal minimum, mean, and maximum daily streamflow values were examined for a set of 516 reference stream gauges located throughout the conterminous United States for the period 1951-2009. Cluster analysis was used to classify the stream gauges into 14 groups based on similarity in their temporal patterns of streamflow. The results indicated that the temporal patterns in flowmetrics (1) have strong spatial coherence within each region, (2) are similar among the three annual flow metrics and the four seasonal flow metrics within each region, (3) indicate some small magnitude trends over time, and (4) are only weakly associated with well-known climate indices. We conclude that most of the temporal variability in flow is unpredictable in terms of relations to climate indices and infer that, for the most part, future changes in flow characteristics cannot be predicted by these indices.