Quality control of Australian sea-level pressure observations

Some methods for the quality control of Australian sea-level pressures from SYNOP messages are described. For the detection of systematic errors (biases) a useful exploratory tool is the calculation of an ‘apparent bias’, defined as the mean over many cases of the observed pressure minus an analysis without the use of the station being checked. Confirmatory tools include time series, and metadata. A check that makes use of a simple property of the Australian pressure reduction method is useful for the detection of pressure reduction inconsistencies. The error rate for such inconsistencies has decreased substantially over the last few years, probably as a result of increased automation. A promising approach to the detection of intermittent errors is the use of station-specific regression equations, in which the predictand is the observed minus short range forecast pressure at the location of the station being checked, and the predictors are observed minus short range forecast pressures at neighbouring station locations.