Homogenization of upper air data utilizing reanalyses
Leopold Haimberger, University of Vienna
Upper air temperature records from radiosondes are known to contain significant shifts that make their use for climate change analysis challenging. In recent years, homogenization methods for radiosonde temperature and wind have been developed at ECMWF and the University of Vienna, which use background forecast time series from climate data assimilation efforts (reanalyses) as aid to find breakpoints in the radiosonde time series. The background forecast time series, interpolated to the observation location, is generally a good predictor for the tested observation series, even at remote places, since the data assimilation system is capable of transporting information from data rich into data sparse areas. Since the background forecasts are always available at daily resolution and the standard deviation of the resulting difference series is quite small, it is possible to detect even relatively small shifts in the difference time series.
The shifts may stem from changes in the radiosonde observation practice, but also from changes in the observation vector assimilated in the reanalyses. Especially changes in the satellite observing system can potentially cause shifts in the background forecast time series. Therefore the adjustment of the radiosonde time series is performed using not only background forecast time series as reference (RAOBCORE RAdiosonde OBservation COrrection using REanalyses) but also carefully constructed composites of homogeneous sub-periods from neighbouring station time series (RICH Radiosonde Innovation Composite Homogenization).
When using ERA-40 or ERA-Interim background forecasts as reference, both methods remove much of the pervasive biases especially in the tropics compared to satellite derived temperature time series from 1979 onward. They seem also robust enough to work well also in the more data sparse pre-satellite era. It is hoped that also data from 1938 onward can be adjusted utilizing surface pressure only reanalyses which have recently become available.
A strategy how to construct a global homogenized radiosonde dataset suitable for ingestion into an extended reanalysis from 1938 onward is presented. Also an outlook is given on the application of the homogenization methods to other datasets such as radiosonde humidity.
SECURITY: If you are coming from outside the NOAA campus, you must stop at the Visitor Center to obtain a vistor badge. Please allow 10 extra minutes for this procedure. If you are a foreign national coming from outside the NOAA campus, please contact Barbara Herrli (202-497-3876) or Joe Barsugli (303-497-6042) for information on access requirements.