Assimilation of ozone data from the Michelson Interferometer for Passive Atmospheric Sounding

Global ozone profiles from the Michelson Interferometer for Passive Atmospheric Sounding (MIPAS) were added to an assimilation system that includes ozone data from the Solar Backscatter Ultraviolet/2 (SBUV/2) instrument. Substantial impacts were found and investigated. MIPAS ozone profiles have a vertical resolution close to 3 km, retrieved from infrared limb emission radiances, with a height-dependent error of 4-8%; they were used over 12 layers, spanning between 60-80 hPa and 0.15-0.25 hPa. As a research instrument, MIPAS provides variable coverage from day to day, with occasional data gaps of several days. The operational SBUV/2 data give regular temporal coverage in sunlit areas, yielding total columns and partial profiles, assimilated for Umkehr layers 3 (126-63 hPa) to 12 (<0.25 hPa), with errors comparable to MIPAS near the ozone maximum but increasing substantially at higher and lower levels. The higher accuracy of MIPAS data leads to substantial improvements in the assimilated ozone below the ozone maximum when compared to accurate in situ (ozonesonde) and space-based (occultation) data; vertical gradients in the lower stratosphere are improved. In the Arctic winter of 2002/03, the ability of MIPAS to sample the vortex interior leads to substantial beneficial impacts. In the tropics, assimilation of MIPAS profiles improves analyses in the lower stratosphere and provides substantially better agreement with occultation data in the middle stratosphere, near the ozone maximum. This offsets an underestimate in the SBUV-only analyses caused by transport errors in the forecast. The global improvements in the assimilation lead to local improvements when compared against individual measured ozone profiles, where vertical structures are determined by height dependence in horizontal advection from diverse regions. The spatio-temporal variability in the assimilation is more realistic when MIPAS data are included. Experiments to examine the impacts of temporal gaps in the MIPAS data reveal a 'memory' of 5-10 days in the transport-dominated lower stratosphere. In the middle stratosphere, the impact of missing MIPAS data is more complex; transport errors cause a large initial impact followed by ozone values typical of SBUV assimilation.