Here we describe a new method for the quantification of a global NO_x budget from image sequences of the GOME instrument on the ERS-2. The focus of this paper is on image processing techniques to separate tropospheric and stratospheric NO₂-colums using normalized convolution with infinite impulse response filters (IIR) to interpolate gaps in the data and average the cloud coverage of the earth, the estimation the NO₂ life time and the determination of regional NO_x source strengths.