Operational assimilation of ozone-sensitive infrared radiances at ECMWF

ECMWF has been assimilating ozone-sensitive infrared (IR/O-3) radiances from AIRS, IASI, and HIRS in its operational system since November 2011. We present a detailed assessment of the value of assimilating these observations, as well as the steps that have been taken to successfully merge the infrared information with that provided by ozone products retrieved from UV sensors. In general, the assimilation of the IR/O-3 radiances improves the agreement between the ozone analyses and independent ozone data from MLS and sondes, especially in the UTLS and at high latitudes in the Southern Hemisphere winter. However, blending the information provided by the IR/O-3 radiances with that provided by UV instruments can lead in certain circumstances to a degradation of the ozone analyses in the region of the ozone maximum, particularly at midlatitudes in the summer hemisphere. It is shown that this problem can be alleviated by providing the assimilation system with height-resolved ozone information. The assessment of one-year long assimilation experiments has shown some additional problems in the ozone analyses related to a slow drift of adaptive observation bias correction over time. Previously, just two uncorrected infrared channels (one from AIRS and one from IASI) were used to stabilise (anchor) the bias correction of all other ozone observations against the influence of model systematic error. However, it has been found that their influence (limited primarily to the UTLS) was not sufficient to constrain the ozone assimilation system as a whole (i.e. the total column) over longer time-scales of several months to a year. This issue has been addressed by using additional anchoring observations to constrain the time evolution of the bias correction.