VARIATION IN STREAM SYSTEM AND LANDSCAPE MORPHOLOGY IN THE U.S. FROM ANALYSIS OF STORAGE VOLUME VS. AREA RELATIONSHIPS OF WATER RESERVOIRS

The National Inventory of Dams (NID) is a database maintained by the U.S. Army Corps of Engineers (USACE) to support dam safety. The NID contains information about nearly 80,000 U.S. dams and their reservoirs and provides 60 fields of information on each dam, including its location, purpose(s) for which it was built, the area and storage of its reservoir at normal retention level, and its drainage area. The NID was selected for study because the information is standardized, undergoes QA/QC checking by USACE, and spans a large range of U.S. geomorphic, geologic, and climate conditions. Dams constructed for purposes other than water containment (e.g., tailings and debris control dams) were not considered in this study.

At any given water level a reservoir's storage volume (V) and area (A) reflect the underlying topography of a stream valley and its surrounding watershed. The analysis presented is based on quantifying the relationship between these quantities for NID dams within a given distance of a point on a geographic grid which spans the lower 48 states. For example, consider dams within 90 km of each point of a grid with an 86 km spacing (33% overlap): 698 grid points are associated with more than 20 dams and 512 points are associated with more than 50 dams. At each point regression techniques are applied to the volumes and areas, and in many cases the relationship is approximated by a power-law: log(V) = log(k) + n log(A); 1 < n < 2. In 382 such cases the range in A exceeds 3 orders of magnitude. Because log(k) and n are strongly correlated (r = -0.96), all important variations in the V-A relationship are explored with reference to n. A map of n shows that the V-A relationship varies regionally within the U.S. Regions with n > 1.35 are associated with the CA and western OR and WA; regions with n > 1.25 include the Rockies and parts of the Great Plains. In some parts of the U.S., particularly in central TX and in GA-SC-NC more complicated conditions exist which lead to a break in V-A scaling and poorly constrained power-law regressions with n < 1. Examples from each region are discussed with reference to simplified landscape models consistent with the V-A scaling.