Abstract. Limb scatter instruments in the UV–vis spectral range have provided long-term global records of stratospheric aerosol
extinction important for climate records and modelling. While comparisons with occultation instruments show generally good
agreement, the source and magnitude of the biases arising from retrieval assumptions, approximations in the radiative
transfer modelling and inversion techniques have not been thoroughly characterized. This paper explores the biases
between SCIAMACHY v1.4, OSIRIS v5.07 and SAGE II v7.00 aerosol extinctions through a series of coincident comparisons as
well as simulation and retrieval studies to investigate the cause and magnitude of the various systematic differences.
The effect of a priori profiles, particle size assumptions, radiative transfer modelling, inversion techniques and the
different satellite datasets are explored. It is found that the assumed a priori profile can have a large effect near
the normalization point, as well as systematic influence at lower altitudes. The error due to particle size assumptions
is relatively small when averaged over a range of scattering angles, but individual errors depend on the particular
scattering angle, particle size and measurement vector definition. Differences due to radiative transfer modelling
introduce differences between the retrieved products of less than 10 % on average, but can introduce vertical
structure. The combination of the different scenario simulations and the application of both algorithms to both datasets
enable the origin of some of the systematic features such as high-altitude differences when compared to SAGE II to be
explained.