How many gas samples should be collected per deployment? Katayanagi and Tokida (in
preparation) analyzed the effect of the number of gas samplings on the precision of the flux
calculation by using a Monte Carlo simulation. They found that increasing the number of
samplings increases the precision of the flux calculation but the degree of improvement is
rather limited (Table 4.5). For example, if errors associated with gas sampling and analysis are
10% (CV), estimated error for the CH4 flux based on 2 samplings is 22.5% (CV) and that based
on 5 samplings is still 19.7%. Here it is noteworthy that the fraction of ‘statistically significant’
fluxes increases markedly (28.3% for n = 3, and 91.5% for n = 5, Table 4.6), although the
actual precision for the estimated fluxes does not change much. These results suggest that
the p value of the slope itself does not provide information regarding the precision if the
number of sample is different.