Determination of the Slow-variation Absorption Structure by Nonlinear Regression Method for Improving the Precision of Retrieved NO_2 Density in Atmosphere

The influence of different slow-variation fitted curves on the retrieve of NO2 density by DOAS is analyzed in this paper.It is discovered that random choice of the fitted curve may bring great error to the result,the greatest error exceed 40% to the numerical simulation.A new method searching for the slowvariation absorption structure through nonlinear regression is displayed,by which random choice of polynomial is avoided.By this method the precision of retrieved NO2 density is improved obviously.Finally,the real NO2 vertical column density(VCD) of Chengdu in May 15th,2006.is retrieved.