GSA 2013: Revisiting watershed drainage density: New considerations for hydrologic prediction

While I’ll be missing the festivities at the 125th anniversary edition of the the Geological Society of America, my able collaborator Sarah Lewis will be presenting our work in a session on “Quaternary Geology and Geomorphology: Past, Present, and Future.” Here’s what she’ll be showing off:

Revisiting watershed drainage density: New considerations for hydrologic prediction

S.L. Lewis, M. Safeeq, A.J. Jefferson, G.E. Grant

Watershed morphometry has long been identified as a major control on the shape and character of the hydrograph. Easily extractable landscape-level metrics have been explored for hydrologic prediction in ungaged watersheds, with varying success. In particular, mean drainage density (stream length/watershed area), which has a strong theoretical relationship to flow, has been both heralded and cast aside as an explanatory variable for hydrograph characteristics. However, previous approaches did not account for the spatial heterogeneity in drainage density within a single watershed. For example, many watersheds in the Oregon Cascades are comprised of both young lava flows with limited drainage networks, subtle peaks and sustained baseflows, and older highly dissected volcanics with steep slopes and flashy hydrographs. A mean drainage density fails to represent this dichotomy.

Here we revisit the long-standing conceptualization of drainage density as a good predictor of flow behavior at the landscape level. We depict drainage density (Dd) heterogeneity as a probability distribution function (pdf) of individual drainage densities within a watershed. Rather than limiting Dd to a single number (mean), we use standard quantitative descriptors of the pdf to explore landscape-level controls on flow regime. Two watersheds with similar mean values may have dramatically different pdfs and therefore exhibit variations in flow dynamics. We assert that some of the inconsistent results applying Dd as a predictive variable may be due to the accuracy with which a mean value can capture the behavior of a drainage network. In watersheds where drainage density is homogeneous, mean Dd may provide a good approximation of drainage behavior, but in watersheds where drainage density is heterogeneous, quantitative descriptors of the pdf can provide additional insight into flow dynamics.