Date of Award

4-2016

Degree Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Agricultural and Biological Engineering

First Advisor

Bernard A. Engel

Committee Chair

Bernard A. Engel

Committee Member 1

Dennis C. Flanagan

Committee Member 2

Margaret Gitau

Committee Member 3

Rao S. Govindaraju

Committee Member 4

Venkatesh M. Merwade

Abstract

A watershed-scale hybrid hydrologic model (Distributed-Clark), which is a lumped conceptual and distributed feature model, was developed to predict spatially distributed short- and long-term rainfall runoff generation and routing using relatively simple methodologies and state-of-the-art spatial data in a GIS environment. In Distributed-Clark, spatially distributed excess rainfall estimated with the SCS curve number method and a GIS-based set of separated unit hydrographs (spatially distributed unit hydrograph) are utilized to calculate a direct runoff flow hydrograph, and time-varied SCS CN values and conditional unit hydrograph approach for different runoff depth-based flow convolution are also used to compute long-term rainfall-runoff flow hydrographs. Spatial data processing and model execution can be performed by Python script tools that were developed in a GIS platform.

Model case studies of short- and long-term hydrologic application for four river watersheds to evaluate performance using spatially distributed (Thiessen polygon and NEXRAD radar-based) precipitation data demonstrate relatively good fit against observed streamflow as well as improved fit in comparison with the outputs of spatially averaged rainfall data simulations as follows: (1) application with 24 single storm events using Thiessen polygon distributed rainfall provided overall statistical results in ENS of 0.84 and R2 of 0.86 (improved ENS by 1.8% and R2 by 2.1% relative to averaged data inputs) for direct runoff, (2) simulation of direct runoff flow for the same storm events using NEXRAD data provided ENS of 0.85 and R2 of 0.89 (increase of ENS by 3.0% and R 2 by 6.0%), and (3) 6-year long-term daily NEXRAD data provided total simulated streamflow statistics of ENS 0.71 and R2 0.72 (increased ENS of 42.0% and R2 of 33.3%). These results also indicate that NEXRAD radar-based data are more appropriate for rainfall-runoff flow predictions than rain gauge observations by capturing spatially distributed rainfall amounts and having fewer missing or erroneous records.

The Distributed-Clark model presented in this research is, therefore, potentially significant to improved implementation of hydrologic simulation, particularly for spatially distributed rainfall-runoff routing using gridded types of quantitative precipitation estimation (QPE) data in a GIS environment, as a relatively simple (few parameter) hydrologic model.

Share

COinS