Parsimonious modeling of hydrologic responses in engineered watersheds: Structural heterogeneity versus functional homogeneity

The central premise of this paper is that extensive modifications of land use and hydrology, coupled with intensive management of watersheds in the Midwestern United States over the past century, have increased the predictability of hydrologic responses, allowing for the use of simpler, minimum-calibration models. In these engineered watersheds, extensive tile-and-ditch networks have increased the effective drainage density and have created bypass flow hydrologic systems that generate "flashy" and "predictable" hydrographs. We propose a simple, threshold-based model, the Threshold-Exceedance-Lagrangian Model (TELM), for predicting event hydrographs. TELM was evaluated by comparing predicted hydrographs with those measured over a 4 year period at the outlet of a mesoscale watershed (Cedar Creek, similar to 700 km(2)) in northeastern Indiana. Application of the Soil-Land Inference Model (SoLIM) indicated that, despite structural heterogeneities (e. g., spatial variability in soil taxonomic mapping units), about 80% of the area of the watershed could be assigned a single value of available soil water storage, which was the primary soil parameter that defined hydrograph response. Hydrograph recession curves for multiple events were described well using an exponential function, with the mean arrival time (t(r)) estimated on the basis of the contributing drainage area (A) and the mean occurrence time (t(h)) of the event hyetograph. Also, functional responses (event hydrographs) at the subwatershed scale could be grouped into just two categories on the basis of only spatial variability in rainfall patterns. TELM, with no parameter calibration, matched the observed hydrographs as well as the widely used SWAT model predictions with calibration. Advantages and limitations of the proposed modeling approach were identified, and needed improvements were discussed.