Abstract

Inter-organisational sampling trials were performed at a reclaimed landfill site in Bedfont, West London, UK, to assess the uncertainty in measured concentrations of landfill gas. Eight participants independently measured concentrations of methane, carbon dioxide and oxygen, each on one occasion, over a two week period in February 1999. The organisers monitored the natural variability of the gas concentration over this sampling period using a control borehole in order to separate the variance that arose as a result of the sampling procedure from that caused by natural geochemical variation. A collaborative trial in sampling showed the extent of agreement between participants' results when they measured gases using the same, nominal protocol and equipment. The protocol was identified as fit for the purpose of estimating the mean concentration of methane, carbon dioxide and oxygen in a single borehole to within 21%, 15% and 321% of their assigned values, respectively (at 95% confidence). A realistic estimate of uncertainty in measurements of concentration from different protocols was estimated by the use of a sampling proficiency test, in which participants collected measurements using a protocol selected on the basis of their own professional judgement. Each participant's performance was compared with the consensus mean for each of the three gases using the z-score assessment proposed by the International Harmonised Protocol. This test highlighted the potentially large uncertainty in measured gas concentrations that can arise (91% relative uncertainty for CH₄ in one borehole), and identified that an improved sampling protocol is required if misclassification of the explosive potential of methane is to be avoided.