Quantitative comparison of methods used to estimate methane emissions from small point sources

Abstract. Recent interest in quantifying trace gas emissions from point sources, such as measuring methane (CH4) emissions from oil and gas wells, has resulted in several methods being used to estimate emissions from sources with emission rates below 200g CH4 hour−1. The choice of measurement approach depends on how close observers can get to the source, the instruments available and the meteorological/micrometeorological conditions. As such, static chambers, dynamic chambers, HiFlow measurements, Gaussian plume modelling and backward Lagrangian stochastic (bLs) models have all been used, but there is no clear understanding of the accuracy or precision of each method. To address this, we copy the experimental design for each of the measurement methods to make single field measurements of a known source, to simulate single measurement field protocol, and then make repeat measurements to generate an understanding of the accuracy and precision of each method. Here, for comparison, we present estimates for the percentage difference between the measured emission and the known emission, A, and the average percentage difference for three repeat measurements, Ar , for emissions of 200 g CH4 h−1. Our results show that, even though the dynamic chamber repeatedly underestimates the emission, it is the most accurate for a single measurement and the accuracy improves with subsequent measurements (A = −11 %, Ar = −10 %). The single HiFlow emission estimate was also an underestimate, however, poor instrument precision resulted in reduced accuracy of emission estimate to becomes less accurate after repeat measurements (A = −16 %, Ar = −18 %). Of the far field methods, the bLs method underestimated emissions both for single and repeat measurements (A = −11 %, Ar = −7 %) while the GP method significantly overestimated the emissions (A = 33 %, Ar = 29 %) despite using the same meteorological and concentration data as input. Additionally, our results show that the accuracy and precision of the emission estimate increases as the flow rate of the source is increased for all methods. To our knowledge this is the first time that methods for measuring CH4 emissions from point sources less than 200 g CH4 h−1 have been quantitively assessed against a known reference source and each other.