HR: 1400h
AN: H53D-05    [Abstracts]
TI: Impact of Uncertainty in Runoff and Routing Processes on the Estimation of Non-parametric Unit Hydrographs for the Cypress Creek Watershed, TX
AU: * Soberaski, J
EM: jsobera@clemson.edu
AF: Clemson University, School of the Environment 340 Brackett Hall, Clemson, SC 29631, United States
AU: Moysey, S
EM: smoysey@clemson.edu
AF: Clemson University, School of the Environment 340 Brackett Hall, Clemson, SC 29631, United States
AU: Bedient, P
EM: bedient@rice.edu
AF: Rice University, Civil & Environmental Engineering 113 Mechanical Lab, Houston, TX 77005, United States
AB: Remote sensing data and GIS tools have opened the door to simplify the parameterization of distributed watershed models. However, decisions about the spatial homogeneity of model parameters should also be based on the actual response of a basin to rainfall. For the last 75 years, hydrologists have relied on the unit hydrograph (UH) as a key tool for analyzing watersheds because its shape is directly related to important attributes of the drainage basin controlling runoff (e.g., topography, land use, soil properties, stream network, etc.). Deconvolution of excess rainfall from direct runoff can provide non-parametric estimates of the UH that capture the effects of sub-basin heterogeneity, thereby making these hydrographs particularly useful tools for comparing and classifying watersheds. Due to the mathematical instability of deconvolution, it is unclear whether meaningful UH estimates can be obtained for the purpose of inter-basin comparisons, particularly when processes controlling excess precipitation and direct runoff within the watershed are uncertain. This study evaluates the sensitivity of non-parametric UH's to uncertainty in watershed properties for six gauged sub-basins of the Cypress Creek Watershed, TX. We have used MATLAB to conduct a rainfall-runoff analysis of the Cypress Creek Watershed, TX over a 17 day period during Tropical Storm Allison in 2001. For the six basins analyzed, discharges for Cypress Creek are available at the outflow of each sub-basin and NEXRAD rainfall data are available throughout the watershed. To determine the direct runoff contributed by each sub-basin, incoming upstream flows were routed by simple advection and then subtracted from the downstream discharge record. Excess precipitation was calculated by applying the Green & Ampt infiltration model to the rainfall record for each basin after accounting for initial abstractions and direct losses due to impervious surfaces. In each step of this procedure, the parameters controlling routing and runoff (i.e., stream velocity, soil conductivity and suction, moisture deficit, watershed impervious area and initial abstraction) were highly uncertain, but could be constrained by a range of likely values (e.g., hydraulic conductivity values spanning soils ranging from sands to clays) or limited to a physically plausible range (e.g., stream velocity). By randomly drawing sets of parameters from within their plausible ranges, we were able to assess the impact of uncertainty on our ability to reproduce consistent UH's. We have found that while the magnitude of the UH can vary substantially for different parameter combinations, primarily due to mass balance considerations, the shape of the UH is largely insensitive to the changes in parameter values. This analysis suggests that the UH's found by deconvolution in the Cypress Creek Watershed are useful tools for performing inter-basin comparisons of rainfall-runoff behavior.
DE: 1804 Catchment
DE: 1834 Human impacts
DE: 1838 Infiltration
DE: 1847 Modeling
DE: 1860 Streamflow
SC: Hydrology [H]
MN: 2007 Joint Assembly