Chemical Data Assimilation for Air Quality Forecasting

Unlike the typical design of data assimilation for numerical weather forecasting, initial value optimisation by chemical data assimilation for air quality simulations were often considered as unessential, as errors in initial values were regarded as of vanishing impact. Rather, air surface interactions, especially emissions are a driving forcing factor, while, at the same time, of insufficient knowledge. This fact necessitates to generalise the air quality data assimilation problem to an inversion problem, which calls for a spatio–temporal, that is, model based technique. The paper presents a four–dimensional variational data assimilation implementation, along with a variety of assimilation examples.