hide
Free keywords:
nitrous oxide, N2O, grassland, model, biosphere/atmosphere interactions, biogeochemical processes
Abstract:
We used the trace gas model DayCent to simulate emissions of nitrous oxide (N2O)
from a urine-affected pasture in New Zealand. The data set for this site contained yearround
daily emissions of nitrification-N2O (N2Onit) and denitrification-N2O (N2Oden),
meteorological data, soil moisture, and at least weekly data on soil ammonium (NH4
+) and
nitrate (NO3 ) content. Evapotranspiration, soil temperature, and most of the soil moisture
data were reasonably well represented. Observed and simulated soil NH4
+ concentrations
agreed well, but DayCent underestimated the NO3 concentrations, due possibly to an
insufficient nitrification rate. Modeled N2O emissions (18.4 kg N2O-N ha 1 yr 1) showed
a similar pattern but exceeded observed emissions (4.4 kg N2O-N ha 1 yr 1) by more
than 3 times. Modeled and observed N2O emissions were dominated by peaks following
N-application and heavy rainfall events and were favored under high soil temperatures.
The contribution of N2Oden was simulated well except for a 4-week period when waterfilled
pore space was overestimated and caused high N2O emissions which accounted
for one third of the simulated annual N2O emissions. N2Onit fluxes were overestimated
with DayCent because they are calculated as a fixed proportion of NH4
+ converted to
NO3 , while the data suggest that significant rates of nitrification can occur without
inducing significant N2O emissions. The comprehensive data set made it possible to
explain discrepancies between modeled and observed values. In-depth model validations
with detailed data sets are essential for a better understanding of the internal model
behavior and for deriving possible model improvements.