Estimations of the NH3 EF based on just one single grab sample, one 24-h period, one 48-h period or one 7-d period exhibited a relatively high chance (>30%) to overestimate the true EF. At first glance it seems striking that generally, with the exception of the single grab sampling strategy in dataset 2, there was a higher chance to overestimate the true EF than to underestimate it.
When taking a closer look at the initial dataset, multiple explanations for this outcome can be found. It can be seen in Fig. 1 that the NH3 emission increased more drastically towards the end of the fattening periods for both housing systems (although this was not as pronounced in all fattening periods). For example, the NH3 emission in fattening period 1 for the conventional housing system (Fig. 1A) increased rather abruptly between approximately day 30 and day 50. When only one 24-h or 48-h sample is taken during that fattening period, there is a higher chance for that sample to be taken in a period with relatively higher NH3 emission since there are more periods with relatively higher NH3 emission as compared to periods with relatively lower NH3 emission. If an EF was be estimated only on the basis of that sample, the estimated EF has, consequently, a higher chance to be larger than the true EF.