Application of Multispectral Imagery to Assessment of a Hydrodynamic Simulation of an Effluent Stream Entering the Clinch River Alfred J. Garrett, John M. Iwine, Amy D. Klng, Thomas K. Evers, Daniel A. Levlne, Clell Ford, and John L. Smyre Abstract The ORR consists of 140 km2 of Federal land located in Oak This study investigates the feasibility of using remote sensing Ridge, Tennessee. Three DOE facilities are located on the Om: systems to estimate and model contaminant transport at known the Oak Ridge National Laboratory (ORNL), the Y-12 Center for hazardous waste sites. We used airborne (Daedalus) imagery Manufacturing, and the Former Gaseous Diffusion Site (K-25 and 3D hydrodynamic simulations to estimate the flow rate of Site]. These facilities were established as part of the Manhattan Poplar Creek as it enters the Clinch River, located on the U.S. Project in the 1940s. Scientific experiments conducted in sup- Department of Energy (DOE)Oak Ridge Reservation (ORR)in port of the Manhattan Project, and Atomic Energy Commission Tennessee. The collection of ground-truth data and the simu- and DOE activities since the Manhattan Project, led to the cre- lations were complicated by the variability of the Clinch River ation of hundreds of individual hazardous waste areas. flow, which we attempted to reproduce in the simulations. The Oak Ridge Environmental Management Program (EM) Comparisons of the Daedalus imagery to images created from was established in 1989 to conduct cleanup activities at the the simulations led to the conclusion that the Clinch River/ ORR. A comprehensive understanding of the location, content, Poplar Creek system shifts back and forth between three distinct and environmental characteristics of a waste area is required flow regimes that have different pollutant transport patterns. before a cleanup alternative can be chosen. Remedial investiga- Results of this research suggest that remote-sensing data com- tions and feasibility studies required by the Comprehensive bined with high-resolution numerical modeling and limited Emergency Response, Compensation, and Liability Act (CER- surface measurements might be able to define pollutant CLA) are conducted to determine the appropriate cleanup strat- transport in large bodies of water as well as methods that rely egy for each group of waste areas. only on more extensive suqace measurements. The Clinch River forms the southern and western bound- aries of the ORR. The area of the Clinch River studied is down- stream of most of the ORR. Poplar Creek, which flows directly lntroductlon through the reservation, empties into the CIinch River near the Project Descrlptlon K-25 Site. In this region, the Clinch River is about 150 m wide This project used airborne (Daedalus)imagery, color photogra- with typical depths of 8 to 10 m along the channel. The study phy, and surface measurements to analyze and model aqueous area also includes two Tennessee Valley Authority (TVA) reser- mixing and sediment transport as Poplar Creek enters the voirs, Melton Hill Lake and Watts Bar Lake. The Clinch River/ Clinch River, located on the U.S. Department of Energy (DOE) Poplar Creek confluence is part of the upper end of Watts Bar Oak Ridge Reservation (ORR)in Tennessee. Analysis and model- Lake. Upstream from the Clinch RiverIPoplar Creek confluence ing of the aqueous mixing process are based on the turbidity is Melton Hill Dam, which controls the flow rates and level of the and thermal differences between the ambient river conditions Clinch River near the ORR (MMES, 1995; SAIC, 1996). and surface water entering the river. Because the Clinch River Surface and groundwater drain from the ORR through a net- is the major integrator of all surface and groundwater contami- work of small tributaries of the Clinch River. Poplar Creek is nation from the ORR, investigators require the delineation of the primary Clinch River sub-basin that drains the area near the inflow mixing zones (spatial extent and temporal variations) in K-25 site. Poplar Creek flows for 8.9 km through the K-25 site order to develop efficient sampling plans to monitor actual con- and empties into the Clinch River 18.1 km downstream of Mel- taminant levels prior to restoration and long-term monitoring. ton Hill Dam. Poplar Creek is about 50 m wide, with water depths ranging up to 7 m. Plate 1is a natural color orthophoto- graphic mosaic which shows the Clinch River/Poplar Creek confluence and the close proximity of Poplar Creek to K-25 facil- A.J. Garrett is with the Savannah River Technology Center, ities. The white area in the Clinch River downstream from the Westinghouse Savannah River Company, Building 773-A, ([email protected]). Poplar Creek plume is sunglint. The flow rate in Poplar Creek Office A-1000, Aiken, SC 29808 was high and the water was very muddy when the color photo- J.M. hine is with Science Applications International Corpora- graphs in Plate 1 were taken. This produced a large, well- tion, 4001 North Fairfax Drive, Suite 725, Arlington, VA 22203. The remaining authors are with the Oak Ridge National Labora- tory, Lockheed Martin Energy Research Corporation, Oak Ridge, TN 37831. D.A. Levine is currently with the International Technology Photogrammetric Engineering & Remote Sensing Corporation, 312 Directors Drive, Knoxville, TN 37923-4799. Vol. 66, NO. 3, March 2000, pp. 329-335. C. Ford is currently with the Highlands Soil And Water Conser- 0099-1112/00/6603-329$3.00/0 vation District, USDAA Natural Resources Conservation Ser- 0 2000 American Society for Photogrammetry vices, 4505 George Blvd., Sebring, FL 33782-5837. and Remote Sensing PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING March 2000 329 Winter discharge at Melton Hill Dam is typically 9500 cfs (270 m3/s),but is up to 19,500 cfs (550 m3/s)for short periods with intervening periods of zero discharge. A representative plot of stage data (water surface elevation) located near the mouth of Poplar Creek is shown in Plate 2. The figure shows sharp increases and decreases in Poplar Creek stage (with a lag of several hours) when the Melton Hill discharge is turned on and off (SAIC, 1996). Previous Research Many researchers have combined remote sensing data, numeri- cal modeling, and in situ measurements to understand the pro- cesses that control transport of pollutants and natural constituents of different surface water systems. Jensen et al. (1989) compared Landsat Thematic Mapper (TM)imagery to modeled salinity and suspended sediment distributions in a large lagoon adjacent to the Bay of Campeche. Their model solved the two-dimensional shallow water equations and mass conservation equations for salinity and Total Suspended Solids (TSS). Jensen et al. (1989) found that an atmospherically cor- rected chromaticity transformation of the TM data accounted for 79 percent of the modeled TSS variance. Nichol(1993) used Plate 1. Natural color ortho-photographic mosaic of the con- TM imagery (the 0.45- to 0.52-pm waveband combined with the fluence of Poplar Creek and the Clinch River. Buildings on 0.63- to 0.69-~mwaveband) to distinguish river plumes with the right are part of K-25. Note the mixing of the muddy high dissolved organic matter from adjacent oceanic waters off- Poplar Creek plume into the clear Clinch River. shore from Southeast Asia. Nichol found that the same river plumes could be identified in the 10.5-to 12.5-pm thermal TM waveband. Nichol used these results to derive imagery-based classes of water quality, which can be related to environmental standards for water use. Roberts et al. (1995) showed that defined turbidity plume in the Clinch River that stayed visible multispectral video imagery and statistical modeling could be further downstream than normal. used to determine TSS concentrations in lakes within Canada's In addition to the transport from the K-25 area, the east fork Mackenzie Delta. Jardine et al. (1993)also used a shallow-water of Poplar Creek carries run-off from the eastern, western, and model to help them interpret Advanced Very High Resolution northern portions of the Y-12 facility. The presence of known Radiometry (AVHRR) thermal imagery of oceanic fronts, upwell- hazardous waste concerns at the Y-12 and K-25 sites raises the ings, large eddies, and jets in the coastal regions offshore from possibility that Poplar Creek could carry contaminants into the northern British Columbia. Jardine et al. (1993) simulated tidal Clinch River (MMES, 1995; SAIC, 1996),which is why it was currents with the numerical model, which helped them to iden- selected for this research. tify the fronts, eddies, and other circulatory features in the Daily fluctuations of the Clinch River water level are AVHRR imagery. Miller and Cruise (1995) used a hydrologic caused by the generation of hydroelectric power at Melton Hill model to simulate runoff and sediment transport from a drain- Dam and alter the natural flow of Poplar Creek. age basin in Puerto Rico to coastal waters that contain coral reefs. They calibrated their model with spatial maps of sus- r pended sediment concentrations derived from Calibrated Air- borne Multispectral Scanner (CAMS) imagery. Miller and Cruise 224.65 (1995) found a highly significant negative correlation between 224.6 the modeled sediment transport and annual growth rates for the coral. Malm and Jonsson (1993) found that a simple energy 224.55 conservation model could be used to model the progression of 400 a thermal bar in Russia's Lake Ladoga, based on AVHRR imagery. 224.5 Finally, Zheng et al. (1993)derived estimates of total mass flow B 224.45 in and out of Delaware Bay and mean tidal velocity at the E 300 t t mouth of the Bay using space shuttle time-series photographs. C 224.4 s e 200 Data Collection 8 m.st This research combined ground-truth data derived from in situ sensors with remote-sensing data to drive hydrodynamic 100 224.9 model simulations of the transport of thermal energy and sus- 224.25 pended sediment via surface water.