Effects of Elevation Errors on the Calculation of Station Pressure

Identifying and correcting elevation errors is essential for ensuring the station/surface pressures used in numerical weather prediction (NWP) are unbiased. Elevation comes into play not only as the vertical coordinate for the station pressure observation, but also in computing station pressure from altimeter setting. This presentation describes several types of elevation errors in terms of their impact on station pressure, provides examples of errors in both elevation and station pressure, and discusses means of detecting and mitigating these errors.

For METAR observations, surface pressure must be calculated from the reported altimeter setting. Since this calculation is dependent on the station elevation, errant elevations will lead to incorrect station pressures. Many of these elevation errors are small (4 m to 10 m) and cause pressure errors of 0.5 mb to 1.5 mb, but large errors that range from 60 m to 500 m can cause pressure errors of up to 50 mb.

SYNOP is the WMO TAC (Traditional Alphanumeric Code) text-based format used for the international exchange of surface observations. In U.S. practice, SYNOP reports are generated from METAR observations and so their surface pressures are converted from altimeter setting. The METAR-to-SYNOP conversion depends on a station table that matches five-digit WMO block-station numbers with corresponding four-character ICAO identifiers and that provides elevations used in the surface pressure calculation. Matching a WMO block-station number with the wrong ICAO identifier can lead to some egregious elevation errors of up to 500 m. The station location list also has errors of its own, including incorrect elevations and typos in the station metadata.

Another source of error results from the conversion of TAC SYNOP reports to the Binary Universal Form for the Representation of meteorological data (BUFR), the WMO standard binary format intended to replace all TAC codes. BUFR surface land reports for U.S. stations are converted from TAC SYNOP reports using a station table to supply the station metadata; this station table is different from the one used in the METAR-to-SYNOP conversion, leading to yet another source of error. Additional elevation errors come from dual-location stations, which are stations for which the surface station and upper-air station are at different locations but share the same ID. In principle, the WMO station database OSCAR/Surface allows two entries for one identifier under these circumstances, but in practice many dual-location stations have only a single entry. Efforts to correct the metadata for upper-air stations have led to this single entry being given the station metadata for the upper-air station, leading to errors when these values are used for a non-collocated surface station. Dual-location stations have been found to have pressure errors ranging from 7 mb to 50 mb.

While it would seem that the egregious elevation errors would cause the most surface pressure biases, errors of this scale are often detected by quality control methods and do not cause biases. The most serious elevation errors are the errors around 10 m since these errors are extremely difficult to detect with quality control methods and they go on to create small but significant surface pressure biases.