Assessing Selenium Contamination in the Irrigated Stream–Aquifer System of the Arkansas River, Colorado

Abstract

Prudent interventions for reducing selenium (Se) in groundwater and streams within an irrigated river valley must be guided by a sound understanding of current field conditions. An emerging picture of the nature of Se contamination within the Lower Arkansas River Valley in Colorado is provided by data from a large number of groundwater and surface water sampling locations within two study regions along the river. Measurements show that dissolved Se concentrations in the river are about double the current Colorado Department of Public Health and Environment (CDPHE) chronic standard of 4.6 microg L(-1) for aquatic habitat in the upstream region and exceed the standard by a factor of 2 to 4 in the downstream region. Groundwater concentrations average about 57.7 microg L(-1) upstream and 33.0 microg L(-1) downstream, indicating a large subsurface source for irrigation-induced dissolution and mobilization of Se loads to the river and its tributaries. Inverse correlation was found between Se concentration and the distance to the closest identified shale in the direction upstream along the principal groundwater flow gradient. The data also exhibited, among other relationships, a moderate to strong correlation between dissolved Se and total dissolved solids in groundwater and surface water, a strong correlation with uranium in groundwater, and power relationships with nitrate in groundwater. The relationship to nitrate, derived primarily from N fertilizers, reveals the degree to which dissolved Se depends on oxidation and inhibited reduction due to denitrification and suggests that there are prospects for reducing dissolved Se through nitrate control. Current and future results from these ongoing studies will help provide a foundation for modeling and for the discovery of best management practices (BMPs) in irrigated agriculture that can diminish Se contamination.

Regional scale selenium loading associated with surface coal mining, Elk Valley, British Columbia, Canada

“…However, these reduction reactions are inhibited in the presence of oxygen or nitrate (Zhang and Moore, 1997). Studies have reported that SeO 4 2− accounts for more than 90% of soluble Se in oxygenated waters (Gates et al, 2009;Gerla et al, 2011). In surface mining areas Se often exists at concentrations in excess of biological needs (Lindberg et al, 2011).…”

Assessing the effectiveness of land and water management practices on nonpoint source nitrate levels in an alluvial stream–aquifer system

“…In addition, overfertilization on cultivated fields has results in elevated groundwater C NO 3 , assessed by samples from 52 observation wells and 15 surface water sites (Arkansas River and tributaries) from 10 sampling events between 2006 and 2009 (Gates et al, 2009). In summary, the mean C NO 3 −N for the samples gathered from the 10 surface water sites along the Arkansas River was 1.53 mg/L and the mean for the samples from tributaries was 1.95 mg/L, as compared to the Colorado interim standard of 2 mg/L (CDPHE, 2012) for total N concentration C NO 3 −N þ À C NO 2 −N þ C NH 4 −N Þ in warm rivers and streams.…”

Estimating spatially-variable rate constants of denitrification in irrigated agricultural groundwater systems using an Ensemble Smoother

“…Similar to the forcing terms in the flow model regime, values were assigned only during the irrigation months of April through October, and were spatially-and temporally-variable, with each cell receiving a unique value differing from every other cell, with new values generated for each week of the irrigation season. Values of C s NO 3 for deep percolation water were selected randomly from a normal distribution with a mean equal to 25.0 g m À3 , corresponding to the mean of C NO 3 measured in a regional assessment of groundwater solutes in the LARV (Gates et al, 2009), and a mean of C s DOC equal to 20.0 g m f À3 based on samples of groundwater solutes in the LARV. Standard deviations for both distributions were set to 1.5, producing a spread of values (e.g., 17.0-33.0 g m À3 f for C s NO 3 ) large enough to take into account differences in N cycling (fertilizer application timing and loading, crop uptake, nitrification) in the root zone between locations, yet within the range expected from land management practices that are assumed to be fairly consistent in a spatial area of 4 km by 2 km.…”

“…Standard deviations for both distributions were set to 1.5, producing a spread of values (e.g., 17.0-33.0 g m À3 f for C s NO 3 ) large enough to take into account differences in N cycling (fertilizer application timing and loading, crop uptake, nitrification) in the root zone between locations, yet within the range expected from land management practices that are assumed to be fairly consistent in a spatial area of 4 km by 2 km. The value of C s NO 3 for canal seepage water was set equal to 8.0 g m À3 f , typical for the LARV (Gates et al, 2009). …”

“…The sources of uncertainty considered here include rate of irrigation deep percolation (i.e., recharge to the water table from the unsaturated zone) in time and space, rate of canal seepage, C NO 3 of both deep percolation water and canal seepage water, dissolved organic carbon (DOC) concentration of deep percolation water, spatially-variable hydraulic conductivity of both the alluvium and bedrock shale material, spatially-variable heterotrophic denitrification rate constants within the alluvium, and, since the LARV is underlain by bedrock shale (Gates et al, 2009), spatially-variable autotrophic denitrification rate constants. Deep percolation, canal seepage, and leaching of NO 3 and DOC are based on field monitoring studies performed in the LARV.…”