Off-line atmospheric chemistry and air quality numerical models are driven by uncertain
forcing fields: emissions, boundary conditions, wind fields, vertical turbulent diffusivity,
kinetic chemical rates, etc. Data assimilation can help assess these parameters or fields of
parameters. Because the parameters are a priori much more uncertain than the fields
diagnosed in meteorology and oceanography, data assimilation is much more an inverse
modelling challenge in this context. In this study we experiment on these ideas
by revisiting the Chernobyl accident dispersion event over Europe. We develop a
fast four-dimensional variational scheme (4D-Var) which seems appropriate for
the retrieval of large parameter fields and for the retrieval of parameters that are
non-linearly related to concentrations. The 4D-Var, and especially an approximate
adjoint of the transport model, are tested and validated using several advection
schemes, quite influential on the forward simulation as well as for the data assimilation
results.
Firstly, the inverse modelling system is applied to the dry deposition and the wet
deposition parameters. It is then applied to the emission field and larger parameter fields, such
as horizontal and vertical diffusivities, or even dry deposition velocity field. The
crucial question of deciding whether the inversions are just tuning of parameters, or
retrieval of physically meaningful quantities is discussed. As a by-product, the
choice of parameters for the Chernobyl dispersion simulation used so far in the
literature is shown to be supported by the study, while the inversion of some of the
parameter fields are shown to improve the skills of the simulation significantly. |