|HOPPE, BRENDALYNN - Oregon State University|
|WHITE, DENIS - Us Environmental Protection Agency (EPA)|
|HARDING, ANNA - Oregon State University|
|Mueller Warrant, George|
|HOPE, BRUCE - Ch2m Hill, Inc (NORTH AMERICA)|
|MAIN, ERIC - Ch2m Hill, Inc (NORTH AMERICA)|
Submitted to: Journal of Water and Health
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 4/13/2014
Publication Date: 6/30/2014
Citation: Hoppe, B.O., White, D., Harding, A.K., Mueller Warrant, G.W., Hope, B.K., Main, E.C. 2014. High resolution modeling of agricultural nitrogen to identify private wells susceptible to nitrate contamination. Journal of Water and Health. 12(4):702-714.
Interpretive Summary: Nitrate from commercial fertilizers and livestock manure is both a critical input in crop production and a potential source of groundwater contamination. Landuse data was combined with Extension Service fertilizer recommendations by crop type to create a raster of nitrate nitrogen input to agricultural fields across Oregon. Statistical regression models were then tested looking for possible links between nitrate fertilizer application rates, soil properties, and private domestic well water contamination reports obtained from public data for real estate transactions over the past decade. The model found that there was a statistically significant positive relationship between nitrate application rates from both fertilizer and manure and nitrate concentrations in well water. However, since the model only captured 15% of the total variation in nitrate levels among wells, it is highly probable that other factors not included in the model were actually more important than fertilization rates in determining whether or not individual wells had too much nitrate for safe use as drinking water.
Technical Abstract: Oregon’s Domestic Well Testing Act (DWTA) links testing to property sales enabling continuous data collection on private water sources. This study investigates use of DWTA data as a sentinel surveillance system for monitoring exposures to well contaminants, particularly nitrate. A land use regression (LUR) model was developed to predict nitrate concentrations in well water using manure (manure-N) and fertilizer nitrogen (fertilizer-N), soil leachability, and groundwater data. High resolution agricultural nitrogen datasets enabled hazard characterization relevant to point locations. Univariate regression analyses were significant for fertilizer-N and soil leachability only. Results from the final LUR model regressing these variables on nitrate were significant. Application of the model to DWTA data demonstrated a positive correlation between predicted and observed nitrate concentrations. However, fertilizer-N and soil leachability explained only 15% of the variance in nitrate, suggesting that model improvements are needed before DWTA data are considered truly valid as “sentinels” of well contamination. Additional variables such as proximity to septic systems, well age or depth may increase overall predictive ability of the model in future iterations. This study demonstrates the need for effectively tracking fertilizer and manure loading to small areas and the importance of this information for characterizing private well water contamination.