Radisson observations since 1958 have been the backbone of the global meteorological observing system, providing three-dimensional information about the state of the atmosphere in terms of temperature, humidity and wind. In order to be useful for detecting changes of the 3D-structure of the atmosphere, the measured data must be temporally homogeneous. This is generally not the case for long time series of station data due to changes of instrument, observing practise and location. It is proposed to develop an automatic homogenisation method to be applied to time series of the global radio- stoned station data with the aim to remove most of the discontinuities. Time series of monthly means of temperature and humidity 1958-2002 based on the now almost complete ERA-40 dataset will be homogenized. The method will be based on statistical analysis of the time series of the differences between the first guess (i.e. 6-hour forecasts of the ERA-40 assimilation system) and radiowomen observations. Due to the quality of the first guess these differences are generally quite small. Thus discontinuities or breaks in the radiowomen time series should be visible and automatically detectable more often than is the case with difference series from other datasets. Together with metadata of the radiowomen observations most in homogeneities should be detected and corrected. The work will be performed at the European Centre for Medium-Range Weather Forecasts, which has produced the ERA-40 dataset using an advanced multivariate variation assimilation system using all available global observations. As an important product an archive of quality-controlled observations with the departures from the first guess and analysis has been produced. The archive also includes the radiowomen data to be used for this study. ECMWF has also developed an automatic temperature bias correction for radiosondes that may be applicable with some modifications.
Call for proposal
See other projects for this call