Repository logo

Analysis of selenium cycling and remediation in Colorado's Lower Arkansas River Valley using field methods and numerical modeling




Romero, Erica C., author
Gates, Timothy K., advisor
Bailey, Ryan T., advisor
Hoag, Dana L. K., committee member

Journal Title

Journal ISSN

Volume Title


Groundwater and surface water concentrations of selenium (Se) threaten aquatic life and livestock as well as exceed regulatory standards in Colorado's Lower Arkansas River Valley (LARV). Se is naturally present in surface shale, weathered shale, and bedrock shale in the region. Excess nitrate (NO3) from irrigated agricultural practices oxidizes Se from seleno-pyrite present in shale and inhibits its chemical reduction to less toxic forms. Irrigation-induced return flows and evapotranspiration induce high concentrations of Se in the alluvial groundwater resulting in substantial nonpoint source loads to the stream system. This research uses three main components to address the need to better describe and find solutions to the problem of Se pollution in the LARV: (1) Se data collection in streams to characterize solute and sediment concentrations, (2) development of a conceptual model of in-stream Se reactions, and (3) application of existing calibrated groundwater models to explore alternative Se remediation strategies. Data in the form of Se solute samples, Se sediment samples, and related water properties were collected during four different sampling events in 2013 and 2014 at several locations in the stream network in an effort to understand the various species of Se and how they cycle through the surface water environment. A conceptual representation of the major chemical reactions of Se in the water column and sediments of streams was described and incorporated into the OTIS (One-Dimensional Transport with Inflow and Storage) computational model of stream reactive transport for future coupling to the MODFLOW-UZF and RT3D-UZF groundwater models. The new version of OTIS, now called OTIS-MULTI, allows for simulation of the cycling of multiple Se species in the river environment. Lastly, five best management practices (BMPs) were tested using MODFLOW-UZF and RT3D-UZF: improved irrigation efficiency (reduced irrigation), lining or sealing of canals to reduce seepage, lease fallowing of irrigated fields, improved fertilizer management (reduced fertilizer), and enhancement of riparian buffers. The impact of each of these BMPs on Se loading to the stream network was evaluated individually over three scenarios in which the adaption of each BMP is incrementally increased. In addition, various combinations of three and four BMPs were simulated and compared. Water samples gathered from the Arkansas River had total dissolved Se concentrations ranging from 6.1 to 32 μg/L (ICP method), compared to the Colorado chronic standard of 4.6 μg/L, while concentrations in samples gathered from tributaries ranged from 6.04 to 29 μg/L (ICP method). The groundwater and drinking water standard from the National Primary Drinking Water Regulations for selenium is 50 μg/L (USEPA, 2016). Concentrations of total Se (sorbed, reduced, and organic) in river bed sediments ranged from 0.16 to 0.36 μg/g with concentrations in river bank samples ranging from 0.26 to 1.78 μg/g. About 70 to 80% of Se in bed and bank sediments was found to be in a reduced or organic form. Analysis also reveals statistically significant high correlations of 0.70-1.00 between sorbed SeO3 (bed sediment (µg/g) and sorbed SeO4 (bed sediment) (µg/g); sorbed SeO3 (bed sediment (µg/g) and estimated precipitated and organic Se (bed sediment) (µg/g); sorbed SeO4 (bed sediment) (µg/g) and estimated precipitated and organic Se (bed sediment) (µg/g); ammonium (mg/L) and nitrite-nitrogen (NO2-N) (mg/L); NO3-N (mg/L) and total dissolved Se (µg/L); and sorbed SeO4 (bank sediment average (µg/g) and an estimate of precipitated and organic Se (bank sediment average) (µg/g). The conceptualization of key Se reactions was incorporated into OTIS-MULTI and must now be tested and calibrated for future application. The groundwater model results indicate that the individual BMP scenarios that most effectively decrease Se total mass loadings to the Arkansas River and its tributaries are: lease fallowing, resulting in a 15% decrease in predicted mass loading; reduced irrigation, with an 11% decrease; canal lining or sealing, with a 10% decrease; enhanced riparian buffer, with a 7% decrease; and reduced fertilizer, with a 3% decrease. In comparison, a BMP combination of lease fallowing, canal lining or sealing, enhanced riparian buffer, and reduced fertilizer was predicted to reduce loads by 46% and a combination of reduced irrigation, canal lining or sealing, enhanced riparian buffer, and reduced fertilizer by 44%. The hope, to be proven by future investigations, is that these reduced loads will contribute to lower concentrations in the river system.


Rights Access


Arkansas River
water quality


Associated Publications