ESTIMATING MEAN WEIGHTED TEMPERATURE OF THE ATMOSPHERE FOR GLOBAL-POSITIONING-SYSTEM APPLICATIONS

A water-vapor-weighted vertically averaged mean temperature of the atm osphere, T-m, is a key parameter in the retrieval of total water conte nt from the measurements of Global Positioning System signal delays. T he accuracy of precipitable water estimates is proportional to the acc uracy of T-m. The geographic and seasonal variability of T-m based on 23 years of radiosonde soundings at 53 globally distributed stations i s presented. Several methods for estimating T-m were evaluated by comp aring the estimates against the actual T-m values. Site-specific clima tology or site-specific linear regression was superior to the geograph ically and seasonally invariant regression relationship typically used to estimate T-m. Relative errors at most stations were less than 2%, which corresponds to absolute errors of precipitable water of 0.1-0.5 mm. The station-specific linear regression was superior to climatologi cal means as an estimator except in the tropics, where correlations be tween T-m and T-sfc were not high, Also, a physical model was develope d to indicate the relationship between T-m and other commonly used atm ospheric variables.