PhD Thesis, 2010 - University College Cork

Gap-filling techniques for the annual sums of nitrous oxide
6.5. Discussion
The presented approaches had similar outcomes. As was expected sequential
methods still had some gaps in the final time series. The same was observed by Leahy
et al. (2004), who observed two gap periods longer than 5 days in the final time series.
In the case of moving-average technique the length and quantity of the remaining gaps
is influenced by different averaging windows and methods based on extrapolation are
likely to always have some gaps at least in the final daily time series.
In 2007, the largest contribution to the remaining gaps was due the down-time in
the beginning of year. Sequential techniques, effectively, failed to alleviate this issue;
therefore, the obtained annual are considered an underestimation of the true annual
flux. As the first three months of a year have typically little to no agricultural activity,
the N 2 O flux is likely to be low; from our observation in other years, it is estimated that
for January–March period total flux amounts to approximately 0.3 kg N 2 O–N ha −1 .
It is notable that performance of the most trivial approach (annual extrapolation) is
very close to the consensus value in the absence of very long gaps. It shows that when
the required precision is about 0.1 kg N 2 O–N ha −1 this method might serve as good
first estimate of the annual flux.
The look-up table was the only approach that considered the environmental
variables of the flux. This method is sensitive to the parameters selected as well as the
separation limits or bin sizes. All the chosen parameters are well known as the main
drivers of the N 2 O emission (Dobbie and Smith 2003b; Dobbie and Smith 2003a). Lag
periods, 12-hour rainfall and 5-day fertiliser applications, were based on our previous
work at this site and reflected precision of the available data. In this study the LUT
58