One of the most important groundwater problems in Oregon, Washington and Idaho is the long-term decline of the groundwater surface level, which has been intensified by wells through discharging water from aquifers. Groundwater contamination from agriculture aggravates the depletion problem in irrigated regions since quality has a decisive role in ways of water use. We developed a spatial agricultural groundwater extraction model by coupling a hydrological model and a contamination migration model. We find that the optimal groundwater extraction is reduced if spatial interactions are incorporated, and that spatial heterogeneities such as crop varieties and soil types affect individual extraction. The socially optimal paths of shadow prices of groundwater quantity and quality depend on time preference, stock effect and dilution effect.