Large scale data assimilation based on RRSQRT-filters; application on atmospheric chemistry models

Abstract

Concentration patterns of air pollutants are usually obtained from either measured concentrations, or from concentrations calculated with an atmospheric model. With the increase of available computing power, one is now able to combine the benefits of both with the aid of data assimilation techniques. The aim of data assimilation is to improve model predictions through incorporation of all available measurement data. A special formulation of the Kalman filter, the so-called RRSQRT-filter, has proven to be an efficient tool for assimilation of data in atmospheric transport models. Two special extensions of the RRSQRT filter have now been made, such that the technique is more robust, and can be applied to a model including non-linear chemistry. The extended filter technique seems to be a useful tool for the assimilation of data with atmospheric chemistry models.