Document Details

A Bayesian approach to infer nitrogen loading rates from crop and land-use types surrounding private wells in the Central Valley, California

Katherine M. Ransom, Andrew M. Bell, Quinn E. Barber, George Kourakos, Thomas Harter | May 7th, 2018


This study is focused on nitrogen loading from a wide variety of crop and land-use types in the Central Valley, California, USA, an intensively farmed region with high agricultural crop diversity. Nitrogen loading rates for several crop types have been measured based on field-scale experiments, and recent research has calculated nitrogen loading rates for crops throughout the Central Valley based on a mass balance approach. However, research is lacking to infer nitrogen loading rates for the broad diversity of crop and land use types directly from groundwater nitrate measurements.

Relating groundwater nitrate measurements to specific crops must account for the uncertainty about and multiplicity in contributing crops (and other land uses) to individual well measurements, and for the variability of nitrogen loading within farms and from farm to farm for the same crop type. In this study, we developed a Bayesian regression model that allowed us to estimate land-use-specific groundwater nitrogen loading rate probability distributions for 15 crop and land-use groups based on a database of recent nitrate measurements from 2149 private wells in the Central Valley.

The water and natural, rice, and alfalfa and pasture groups had the lowest median estimated nitrogen loading rates, each with a median estimate below 5 kg N ha -1 yr -1. Confined animal feeding operations (dairies) and citrus and subtropical crops had the greatest median estimated nitrogen loading rates at approximately 269 and 65 kg N ha -1 yr-1, respectively. In general, our probability-based estimates compare favorably with previous direct measurements and with mass-balance-based estimates of nitrogen loading. Nitrogen mass-balance-based estimates are larger than our groundwater nitrate derived estimates for manured and non-manured forage, nuts, cotton, tree fruit, and rice crops. These discrepancies are thought to be due to groundwater age mixing, dilution from infiltrating river water, or denitrification between the time when nitrogen leaves the root zone (point of reference for mass-balance derived loading) and the time and location of groundwater measurement.

Keywords

Central Valley, groundwater contamination, Groundwater Exchange, modeling, nitrates, water quality