Abstract:

Two operational analyses, one from the U.S. National Centers for Environmental Prediction (NCEP) and the other from the European Centre for Medium-Range Weather Forecasts (ECMWF), provide, among other quantities, atmospheric pressure at the Earth’s surface. Because these analyses are derived from combining observational data with forecast fields from atmospheric models, their characteristics depend upon the type of observation available, the assimilation technique, and the characteristics of the model itself, all of which have changed over the course of time. To assess error levels in analyses of surface pressure, we compare their values to surface barometric observations; to do so, the former are interpolated spatially and temporally to the observation location and time. Over land, surface observations are typically taken at fixed stations, but those over the ocean are determined from a special set of ship-based measurements. The largest errors are found over the high latitudes of both hemispheres, particularly over Antarctica. During our study period of 2000-2003 we observe some long-term regional biases between the analysis fields and the observations, sometimes up to 4 hPa over the southern oceans and high altitude regions of the continents. Maps of total root mean square differences indicate that larger errors occur on all time scales besides the bias. Considering the mean differences between the NCEP and ECMWF values, we note that they have generally decreased in time. Variances of this difference are separated into temporal bands as well for the difference between analyses. For variances on periods roughly greater than 30-days, similar to the time scale resolved by GRACE, we note that the largest differences are over the Antarctic Ice sheet, oceans in the high southern latitudes particularly near 120 W – 180, the Himalayan region, and Greenland.