Simulated effects of hydrologic, water quality, and land-use changes of the Lake Maumelle watershed, Arkansas, 2004–10

Abstract

Lake Maumelle, located in central Arkansas northwest of the cities of Little Rock and North Little Rock, is one of two principal drinking-water supplies for the Little Rock, and North Little Rock, Arkansas, metropolitan areas. Lake Maumelle and the Maumelle River (its primary tributary) are more pristine than most other reservoirs and streams in the region with 80 percent of the land area in the entire watershed being forested. However, as the Lake Maumelle watershed becomes increasingly more urbanized and timber harvesting becomes more extensive, concerns about the sustainability of the quality of the water supply also have increased. Two hydrodynamic and water-quality models were developed to examine the hydrology and water quality in the Lake Maumelle watershed and changes that might occur as the watershed becomes more urbanized and timber harvesting becomes more extensive. A Hydrologic Simulation Program–FORTRAN watershed model was developed using continuous streamflow and discreet suspended-sediment and water-quality data collected from January 2004 through 2010. A CE–QUAL–W2 model was developed to simulate reservoir hydrodynamics and selected water-quality characteristics using the simulated output from the Hydrologic Simulation Program–FORTRAN model from January 2004 through 2010. The calibrated Hydrologic Simulation Program–FORTRAN model and the calibrated CE–QUAL–W2 model were developed to simulate three land-use scenarios and to examine the potential effects of these land-use changes, as defined in the model, on the water quality of Lake Maumelle during the 2004 through 2010 simulation period. These scenarios included a scenario that simulated conversion of most land in the watershed to forest (scenario 1), a scenario that simulated conversion of potentially developable land to low-intensity urban land use in part of the watershed (scenario 2), and a scenario that simulated timber harvest in part of the watershed (scenario 3). Simulated land-use changes for scenarios 1 and 3 resulted in little (generally less than 10 percent) overall effect on the simulated water quality in the Hydrologic Simulation Program–FORTRAN model. The land-use change of scenario 2 affected subwatersheds that include Bringle, Reece, and Yount Creek tributaries and most other subwatersheds that drain into the northern side of Lake Maumelle; large percent increases in loading rates (generally between 10 and 25 percent) included dissolved nitrite plus nitrate nitrogen, dissolved orthophosphate, total phosphorus, suspended sediment, dissolved ammonia nitrogen, total organic carbon, and fecal coliform bacteria. For scenario 1, the simulated changes in nutrient, suspended sediment, and total organic carbon loads from the Hydrologic Simulation Program–FORTRAN model resulted in very slight (generally less than 10 percent) changes in simulated water quality for Lake Maumelle, relative to the baseline condition. Following lake mixing in the falls of 2006 and 2007, phosphorus and nitrogen concentrations were higher than the baseline condition and chlorophyll a responded accordingly. The increased nutrient and chlorophyll a concentrations in late October and into 2007 were enough to increase concentrations, on average, for the entire simulation period (2004–10). For scenario 2, the simulated changes in nutrient, suspended sediment, total organic carbon, and fecal coliform bacteria loads from the Lake Maumelle watershed resulted in slight changes in simulated water quality for Lake Maumelle, relative to the baseline condition (total nitrogen decreased by 0.01 milligram per liter; dissolved orthophosphate increased by 0.001 milligram per liter; chlorophyll a decreased by 0.1 microgram per liter). The differences in these concentrations are approximately an order of magnitude less than the error between measured and simulated concentrations in the baseline model. During the driest summer in the simulation period (2006), phosphorus and nitrogen concentrations were lower than the baseline condition and chlorophyll a concentrations decreased during the same summer season. The decrease in nitrogen and chlorophyll a concentrations during the dry summer in 2006 was enough to decrease concentrations of these constituents very slightly, on average, for the entire simulation period (2004–10). For scenario 3, the changes in simulated nutrient, suspended sediment, total organic carbon, and fecal coliform bacteria loads from Lake Maumelle watershed resulted in very slight changes in simulated water quality within Lake Maumelle, relative to the baseline condition, for most of the reservoir. Among the implications of the results of the modeling described in this report are those related to scale in both space and time. Spatial scales include limited size and location of land-use changes, their effects on loading rates, and resultant effects on water quality of Lake Maumelle. Temporally, the magnitude of the water-quality changes simulated by the land-use change scenarios over the 7-year period (2004–10) are not necessarily indicative of the changes that could be expected to occur with similar land-use changes persisting over a 20-, 30-, or 40- year period, for example. These implications should be tempered by realization of the described model limitations. The Hydrologic Simulation Program–FORTRAN watershed model was calibrated to streamflow and water-quality data from five streamflow-gaging stations, and in general, these stations characterize a range of subwatershed areas with varying land-use types. The CE–QUAL–W2 reservoir model was calibrated to water-quality data collected during January 2004 through December 2010 at three reservoir stations, representing the upper, middle, and lower sections of the reservoir. In general, the baseline simulation for the Hydrologic Simulation Program–FORTRAN and the CE–QUAL–W2 models matched reasonably well to the measured data. Simulated and measured suspended-sediment concentrations during periods of base flow (streamflows not substantially influenced by runoff) agree reasonably well for Maumelle River at Williams Junction, the station representing the upper end of the watershed (with differences—simulated minus measured value—generally ranging from -15 to 41 milligrams per liter, and percent difference—relative to the measured value—ranging from -99 to 182 percent) and Maumelle River near Wye, the station just above the reservoir at the lower end (differences generally ranging from -20 to 22 milligrams per liter, and percent difference ranging from -100 to 194 percent). In general, water temperature and dissolved-oxygen concentration simulations followed measured seasonal trends for all stations with the largest differences occurring during periods of lowest temperatures or during the periods of lowest measured dissolved-oxygen concentrations. For the CE–QUAL–W2 model, simulated vertical distributions of water temperatures and dissolved-oxygen concentrations agreed with measured vertical distributions over time, even for the most complex water-temperature profiles. Considering the oligotrophic-mesotrophic (low to intermediate primary productivity and associated low nutrient concentrations) condition of Lake Maumelle, simulated algae, phosphorus, and nitrogen concentrations compared well with generally low measured concentrations.