THE TOTAL MASS OF THE ATMOSPHERE

Accurate but approximate formulae for determining the mass of the atmo sphere in terms of the surface pressure p(s) are derived and applied t o globally analyzed data from the European Centre for Medium-Range Wea ther Forecasts (ECMWF) for 1985 through 1993. The formulae take into a ccount effects of the shape of the Earth and variations in gravity wit h latitude and height. Variations in total mass occur because of chang es in the water vapor loading of the atmosphere. Independent computati ons are made of the surface pressure due to water vapor p,, which is p roportional to the precipitable water, using the ECMWF analyses of spe cific humidity. Spurious trends in both the mass of dry air and the at mospheric moisture are found to arise from changes in the analysis sys tem at ECMWF, confounding attempts to seek real trends associated with climate change. For the recent 4-year period 1990 to 1993 the mean an nual p(s) was 984.76 mbar with a maximum in July of 984.98 mbar and a minimum in December of 984.61 mbar which correspond to a total mean ma ss of the atmosphere of 5.1441 x 10(18) kg with a range of 1.93 x 10(1 5) kg throughout the year associated with changes in water vapor in th e atmosphere. The global mean p(w) for 1985-1993 is 2.58 mbar, but val ues are 5 to 10% lower after mid-1992. Using the Special Sensor Microw ave Imager data to make adjustments, the best estimate of the annual g lobal p(w) is 2.4 mbar, corresponding to similar to 2.5 cm of precipit able water. The total atmospheric moisture as given by p(w) varies wit h an annual cycle range of 0.36 mbar, a maximum in July, and a minimum in December. Thus the mean mass of water vapor is 1.25 x 10(16) kg an d the dry air mass is 5.132 x 10(18) kg, corresponding to a mean surfa ce pressure of 982.4 mbar. Overall uncertainties are similar to 0.1 mb ar or 0.5 x 10(15) kg in total mass and about double those values for atmospheric moisture content. As well as the global means, hemispheric mean values and meridional profiles of p(s) and p(w) are presented fo r the mean annual cycle and as latitude-time series to show the intera nnual and longer-term variability.