Abstract
Reduction of station pressure to mean sea level (MSL) is a difficult procedure. In general, the temperature structure of the fictitious air column between station height and MSL is not known and has to be estimated somehow. Normally, station pressure is reduced to MSL only for stations with relatively low elevations above sea level (ASL). At higher stations, station pressure is usually converted to the height of the closest standard pressure surface.
In the United States, however, station pressure is reduced to MSL for stations as high as 2000 m ASL. In order to reduce the amplitude of the annual MSL pressure variation at stations situated above 305 m ASL (hereinafter referred to as “plateau stations”), a so-called plateau correction is applied at these stations. The correction increases reduced MSL pressure when the actual temperature at the station is greater than the yearly mean temperature at the same station, and vice versa. The correction can therefore change both magnitude and direction of MSL pressure gradients. This is illustrated by means of the average monthly MSL pressure differences between the two cities of Yuma (southwestern Arizona) and Las Vegas (Nevada). Reduced MSL pressure values from plateau stations are used operationally in producing MSL pressure charts.
Similar methods for the reduction of station pressure to MSL are used in the postprocessing procedure of numerical atmospheric models, in order to obtain pressure or geopotential fields below the lowest level of the numerical model. Temperatures, on the other hand, are normally extrapolated from the lowest levels of the numerical model by means of a standard-atmosphere temperature lapse rate. For this reason, fields below the model's orography can be out of hydrostatic balance. This was found to be the case for the elevated regions of the western United States, where the lowest level of a global atmospheric model is usually at a height of around 1500 m ASL.
Nine days of measurements from a part of the lower Colorado River valley are used to evaluate such fields over the southwestern United States during summer. Mesoscale model simulations were carried out using fields from the NCEP–NCAR reanalysis system as basic-state conditions. Model-predicted winds were then compared to measured winds in that part of the lower Colorado River valley, situated approximately 100 km to the south-southeast of Las Vegas. The results showed that, in the lowest 1000 m ASL or so, model-predicted winds within the valley agreed far better with observed winds, when input geopotential fields were hydrostatically recalculated below 850 hPa before using them as basic-state conditions in the mesoscale model.
Ten years of geopotential fields were hydrostatically recalculated below 850 hPa. The hydrostatically recalculated 1000-hPa geopotential fields for summer show an average position of the thermal low that is about 450 km to the north and somewhat to the east, compared to the position in the original NCEP–NCAR 1000-hPa summer geopotential fields. In addition, the thermal low is about 40 gpm (≈5 hPa) deeper in the recalculated 1000-hPa geopotential fields. During winter, however, differences between hydrostatically recalculated 1000-hPa geopotential fields and original NCEP–NCAR 1000-hPa geopotential fields were very small.
Corresponding author address: Matthias Mohr, Renewable Energy Systems, Ltd., 7000 Academy Park, Glasgow G51 1PR, United Kingdom. Email: matthias.mohr@res-ltd.com
