The behavior of a numerical solute transport model is examined under calibration conditions. Predicting ground water remediation using pumping wells is also examined. Model calculations are compared for calibration conditions and remedial conditions to determine the reliability of parameter estimates obtained from model calibration. Parameters investigated include hydraulic conductivity, aerobic biodegradation, retardation, and source terms. Multiple criteria including RMS errors, plume area, and dissolved mass are used to evaluate model output. The results of the sensitivity analysis are applied to a modeling study and remedial design at a Superfund site. The model is calibrated to historically observed data using two distinct data sets to show the non-uniqueness of the solution. A remedial pumping system is devised to remove contaminated water from the aquifer, and the performance of the system is examined under variations in model parameters. The effects on the remedial system of a residual source are also examined.

scFLOTRAN, a code written by a Rice faculty member, is designed to solve the advection-dispersion-reaction equation for contaminant transport in three dimensions. In this thesis, scFLOTRAN is tested in several different scenarios to determine the accuracy of the code and to eliminate programming errors. The testing scenarios utilized include testing in radially dominated flow conditions, testing against analytical solutions, and testing against a three dimensional model of the Borden Landfill site. Overall, the testing was successful. In the tests with analytical solutions, scFLOTRAN matches the analytical solutions, generally within a few percent, and often within one percent, as measured by the average absolute error. In the testing against the Borden Landfill model, scFLOTRAN performed well, demonstrating the usefulness of its variable grid thickness option. As a final test, and as an exercise in itself, scFLOTRAN is also used to model trichloroethylene (TCE) contamination at a site in Arizona. TCE was discovered in two City of Phoenix water wells in late 1981, a discovery which led to placing the area on the National Priorities List, and extensive study of the area. The emphasis of the modeling is to determine conditions in the subsurface before the discovery of TCE in the water wells. Of the several possible sources of contamination in the area, modeling suggests that one in particular is responsible for the contamination in the city water wells prior to 1981.

Advanced radar (NEXRAD) was used for collection of rainfall data over Brays Bayou in Houston, TX for the storms in October 1994 and January 1998. The radar rainfall data was calibrated to the rain gauge system within the watershed and used as input into the hydrologic model, HEC-1. For the October 1994 storm, the hydrologic model was calibrated to produce a predicted hydrograph with a peak flow within 6% and a time-to-peak within one-half hour of the measured storm hydrograph at Main Street. The calibrated model was used to create an "Isoflow" chart, rainfall intensity versus duration, for peak-flow values at Main Street. This plot was proven successful using rainfall data from historical storms and will serve as a more accurate flood prediction tool to provide advanced flood warning to the Brays Bayou Watershed.

Two models were incorporated to evaluate the hydrologic and hydraulic behavior of the Harris Gully watershed in Houston, Texas. A HEC-1 model was created and calibrated to three historical storms for average errors of 6% in peak flow, 4% in volume and 0.9 hours in peak time. The StormCAD model was used to determine capacity of the storm sewer system experiencing surcharged conditions and backwater effects. These capacities were incorporated into the HEC-1 model as diversions from the computed total runoff. The resulting model predicts the volume and approximate location of overland flow that can flood the system for any rainfall event scenario. Relationships were explored among storm intensity, duration, tailwater elevation and the resulting overland flow volume. Results show that any tailwater elevation over 40 feet shuts down the drainage capacity of the storm sewers. Three detention pond alternatives are proposed and their impacts were evaluated for flood-relief potential.

A controlled pilot scale aquifer system known as the Experimental Controlled Release System (ECRS) was used to evaluate the bioremediation of PCE DNAPL source zones. The addition of DNAPL in the ECRS made a thorough hydrologic evaluation of the system both more essential and problematic. The focus of this research was to develop a detailed site characterization to help ensure the effective delivery of electron donor (Hydrogen Releasing Compound) and microbes for dechlorination of PCE. Transient water level monitoring, tracer tests, MODFLOW models and spatial analyses of aqueous samples were used to define the hydrogeology of the ECRS. Results of these experiments found that addition of the above constituents resulted in no overall aquifer clogging or impacts on the hydraulics. Additional findings of this investigation contradict some previous laboratory studies with regard to the influence of methanogens on dechlorination and methane production, which is believed to result in aquifer clogging.

The primary objective of this project was to develop hydrologic and hydraulic models for the floodplain analysis for Brownsville, Texas and analyze mitigation alternatives within a Geographic Information Systems (GIS) and digital data framework. This framework allowed for flexible manipulation of data and easy interpretation of the results. A methodology was developed that would make improvements to the more standard floodplain analysis approach with respect to both cost-effectiveness and accuracy. The project relied extensively on readily available digital data with a minimum, but sufficient, incorporation of field data for model verification purposes. Examples of this type of data included a LIDAR digital elevation model, global positioning system data, aerial photographs, and land use and land cover data. Another project objective included the collection and organization of various hydrologic and hydraulic data sources within an ArcView GIS system. The consolidation of this data will allow for rapid and efficient model updates as required in the future, and for the analysis of various alternatives for future flood control.

Flood Warning Systems are not a new commodity to the hydrologic world, however, the advancements in GIS data, hydrologic modeling tools, NEXRAD, and internet capabilities have allowed for improvements in flood prediction. This study applied techniques originally used in FAS-1 for a watershed in Houston, Texas to a large, highly sloped watershed in Austin, Texas where aquifer recharge is prominent. GIS was helpful in the framework of the model producing valuable hydrologic parameters and delineating the watershed. HEC-1 was used to create a model as the basis for a real-time radar based Flood Alert System. Advantages to the HEC-1 model are its ability to run in real time and the addition of gage-adjusted radar to the input. The HEC-1 model was tested and calibrated to various historical storms and is ready for real-time application. Together, these entities provide a strong foundation for the Flood Alert System in Austin, Texas.

The harmonic mean, 7Q2 (7-day, 2-year low-flow), and 7Q10 are streamflow characteristics used to determine stream waste load allocations. Predictive equations for these characteristics of ungaged Texas streams are developed using multiple regression analysis to relate streamflow characteristics to basin parameters at 63 gaged streams. Gamma, LPIII (Log-Pearson type III), and mixed-LPIII distributions are used to predict the 7Q2's and 7Q10's. Basin parameters used in the analysis include contributing drainage area, channel length and slope, basin shape factor, precipitation, and hydrologic soil group. The harmonic mean regression equation included slope, precipitation, and area and had a higher adjusted R$\sp2$ than the 7Q2 and 7Q10 equations. Gamma 7Q2 and 7Q10 estimates yielded more reliable equations than LPIII and mixed-LPIII estimates. A region-based regression analysis yielded an even higher adjusted R$\sp2$ for the harmonic mean than the statewide equation, implying that regional regression equations may be more reliable than statewide equations for Texas.

Floods are calamities with potential for many fatalities and enormous economic losses. However, the degree of vulnerability to a disaster is a function of human action and behavior. The rapid development of modern metropolitan areas has changed the response of floodplains to rainfall and increased the risk of flooding. The current study is concerned with the flooding along Brays Bayou, a 31-mile waterway located on the southwest part of the city of Houston. The first objective of this study is to simulate the runoff from Tropical Storm Allison and evaluate the use of NEXRAD (next generation radar) for Allison over Brays. T.S. Allison's severity was not identical for all Brays Bayou. Thus, the second objective of this research is to develop scenarios that investigate the significance of the dissimilarity of Allison's rainfall patterns along the watershed to the magnitude of consequences on the middle and lower part of Brays Bayou.

The deployment of the WSR-88D (NEXRAD) radar by the National Weather Service in 1992 has created the possibility for enhancing flood warning systems. Due to the improved spatial resolution of the radar, a better hydrologic prediction is possible on a real-time basis. Rainfall rates can currently be derived from radar data real-time at a resolution of 16 km2, whereas post-processing of the radar data can yield rainfall rates at a resolution a small as 1 km2. This study examined the scale effects of the radar-rainfall input on the hydrologic prediction at two urban watersheds during two recent severe storms and compared the results the hydrologic prediction using rain gages and to the observed hydrograph. Small deviations were observed with the different grid scales, but these differences were considered minor compared to other unknowns in hydrologic modeling. Additionally, lead-times were quantified for both bayous.