Nutrient pollution is becoming a most widespread challenging environmental problem with increasing agriculture activities. A high level of nitrite pollutant is harmful to human and animals, especially to infants and pregnant women. Therefore, it is of paramount importance to develop a field-deployable sensor capable of conducting routine nitrite detection in the environment. Traditionally, the Griess reagent is used to treat a nitrite-containing sample and transduce a red pink color. However, the Griess test is vulnerable to false positives in colored sample matrices and limited in its sensitivity. Herein, we have developed a surface-enhanced Raman scattering (SERS) nanostructured device based on a modified Griess test, where nitrite is detected by transducing a SERS signal. The SERS sensor is built based on an optimized electromagnetic field coupling using finite difference time domain (FDTD) simulations. It is demonstrated for nitrite detection in Monongahela River water sample, achieving a linear detection range from down to part-per-trillion up to part-per-million levels.