The RTTOV-9 upgrade for clear-sky radiance assimilation in the IFS

TitleThe RTTOV-9 upgrade for clear-sky radiance assimilation in the IFS
Publication TypeMiscellaneous
Year of Publication2009
AuthorsBormann, N, Salmond, D, Matricardi, M, Geer, A, Hamrud, M
Secondary TitleTechnical Memorandum
Date PublishedMarch
Type of WorkTechnical Memorandum

The memorandum describes the impact of the upgrade to RTTOV-9 on the assimilation of clear-sky radiances in the ECMWF system. The upgrade includes the use of the new linear-in-tau parameterisation of the source function, the inclusion of the effect of variable zenith angles with height, the use of the new internal interpolation provided by RTTOV-9, and an upgrade to the coefficient files used for RTTOV. The latter includes a move towards kCARTA-based coefficients for infrared sensors. A detailed analysis of the components of the upgrade shows that the changes primarily alter bias characteristics of the assimilated radiances, prompting different responses of the air-mass dependent bias correction. One of the largest changes is due to the move to kCARTA-based RTTOV coefficients which leads to a significant reduction of the absolute size of the corrections for HIRS. After bias correction, most departure statistics for assimilated radiances are overall largely unaltered. Exceptions are reductions in the size of First Guess departures for lower tropospheric HIRS channels resulting from the kCARTA coefficients, and, for specific periods, reductions in the size of First Guess departures for some high-peaking IASI channels which benefit from improvements towards the model top, related to the use of the RTTOV internal interpolation. The use of the RTTOV internal interpolation successfully avoids spikes previously seen in gradients from the radiance data on model profiles. The forecast impact of the upgrade is overall neutral. Mean temperature analyses close to the forecast model's top are considerably altered, as a result of an improved handling of temperature information near the forecast model's top.