Masters Theses

Date of Award

8-1997

Degree Type

Thesis

Degree Name

Master of Science

Major

Environmental Engineering

Major Professor

Bruce Tschantz

Committee Members

William A. Miller, James L. Smoot

Abstract

Low-level radioactive wastes have been generated and buried in trenches at Oak Ridge National Laboratory (ORNL) since 1943 (Tucci, 1992). Groundwater flow through these waste trenches has resulted in the transport and deposition of contaminants to White Oak Creek and its tributaries. The levels of contamination in the creek bed have led to concerns about erosion and off-site transport of contaminated sediment during extreme streamflows.

A streamflow having a one percent probability of being met or exceeded in a given year (100-year return period) was needed for the White Oak Creek Watershed at White Oak Dam, near Oak Ridge, Tennessee, in order to understand how contaminant- bound sediments are mobilized and transported and to assist in identifying, selecting, and designing appropriate remedial alternatives that might be taken to ensure their continued control. The principal existing sediment control structures consist of White Oak Dam, built in 1943, and renovated in 1983, which impounds White Oak Lake, and a permeable gabion dam constructed in 1992 just downstream from White Oak Dam at the mouth of White Oak Creek (Frederick et al. 1996).

When an adequate period of streamflow record is available over homogeneous and stationary watershed conditions, a common approach to determine peak flow return periods is to fit a probabilistic distribution to a series of one peak streamflow per each available year of record. In this case, sporadic flow measurements and the aforementioned dam construction and renovation have caused uncertainties and interruptions in the streamflow record over time. In any event, only five continuous years (1990-94) of streamflow record were available, and another approach had to be undertaken for analyzing and predicting low probability flood frequency.

The approach taken was to simulate a longer streamflow record using a deterministic hydrologic watershed computer model. The model was calibrated using the five year period when streamflow and other hydrologic data were available. The Environmental Protection Agency's Hydrologic Simulation Program--FORTRAN (HSPF) was chosen for the model. 1990-94 data from six rainfall gages were used to develop, calibrate, and validate the model, and a continuous 41-year streamflow record was simulated using the 1953-93 record of precipitation and evaporation available at a single site outside the watershed in nearby Oak Ridge. Annual peak streamflows were extracted from this 41-year simulated record and used for estimating the probability distribution. Attempts were made to identify and to statistically and mathematically account for the random and systematic errors associated with this simulated streamflow record and the flood frequency estimated from it.

Many possible sources of random and systematic error may exist in the estimation of a flood frequency for a watershed. The usual errors encountered with gaged data are only compounded with the use of simulated data. Despite this, useful estimates can be made if the errors are rationally accounted for. Attempts were made to identify and account for each source of error encountered in the simulated streamflow data necessary for the estimation of the flood frequency. After the errors were accounted for, an overall indication of the uncertainty of the resultant flood frequency was estimated by calculating variances reflecting the random frequency analysis and random simulation model errors, assuming them to be independent, combining them, and expressing them as a 90 percent confidence interval.

Download
Files over 3MB may be slow to open. For best results, right-click and select "save as..."

Share

COinS