BAROCLIM
| Project name: | BAROCLIM - Bending Angle Radio Occultation Climatology: Further development of BAROCLIM and implementation in ROPP |
| Project leader: | Barbara Scherllin-Pirscher |
| Partners: | S. Syndergaard, Danish Meteorological Institute (DMI), Copenhagen, DK |
Sponsor: | EUMETSAT ROM SAF CDOP-2 Visiting Scientist Funds |
Abstract:
Due to beneficial characteristics of RO data (e.g., high vertical resolution, high accuracy and precision, or long-term stability), in the UTLS region, these data are often used in atmospheric and climate research. Above the middle stratosphere, however, individual RO measurements suffer from low SNR, which strongly limits the measurements’ quality. When averaging over a large number of profiles, statistical data noise can be reduced, which enables to use data higher up.
In a first study, Foelsche and Scherllin-Pirscher (2012) calculated mean RO bending angles from the F3C satellite constellation and showed that these data are of very high quality at least up to an impact altitude of 60 km. Above that altitude, mean bending angles are increasingly affected by residual data noise, which even yields negative mean bending angle profiles above 80 km. Since Foelsche and Scherllin-Pirscher (2012) only used closed-loop F3C profiles, their mean bending angle profiles do not reach below approximately 8 km.
In this study, I use their mean RO bending angles, extend them with background information at low (below 10 km) and high altitudes (above 80 km) to generate a BAROCLIM spectral model. To extend mean RO profiles above the Mesopause, I perform statistical optimization from 60 km to 80 km using a best-fitting MSIS profile extracted at a specific latitude, longitude, and month. To extend mean RO profiles down to the surface, I apply a cosine transition from 10 km to 15 km using another best-fitting MSIS profile extracted at a specific latitude, longitude, and month.
The BAROCLIM spectral model is expanded into Chebychev polynomials represented by 128 coefficients and zonal harmonics represented by 18 coefficients. The evaluation of the BAROCLIM spectral model with ECMWF rather reveals deficiencies in ECMWF than in BAROCLIM. While differences are small (<0.5 %) below approximately 35 km, larger differences above 40 km are well known biases in ECMWF analyses, which have also been found in comparisons with other satellite data.
A first, very promising, application of BAROCLIM for bending angle initialization and its validation relative to ECMWF clearly shows the potential of BAROCLIM to be used in RO retrieval algorithms.
