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