Data filtering for inverse dispersion emission calculations

Inverse dispersion techniques are used to infer the emission rate of gas sources from concentration measurements and dispersion model calculations. Criteria for the selection of measurement intervals having wind conditions conducive to technique accuracy are examined on the basis of a short range tracer experiment. By introducing a supplementary condition that the measured vertical temperature gradient be quantitatively compatible with Monin-Obukhov similarity theory, it was possible to use a less stringent threshold for the friction velocity than has previous been used (u* ≥ 0.05 m s−1 instead of ≥0.15 m s−1). Under the new criteria a larger proportion of measurement intervals are retained (76% versus 49%), while the ratio of inferred to actual emission rate QLS/Q exhibits negligible bias (average QLS/Q = 1.00) and an acceptably small level of random error (interval-to-interval standard deviation σQ/Q = 0.25).