Assuming that state-of-the-art air quality models are accurate, then the precision and accuracy of their results directly depend on the precision and accuracy of their geographical, meteorological and emission input data. There are important applications, such as open pit mining, in which emission data are the main source of uncertainty. In such cases, historical air quality experimental data are typically available. The present work proposes a backward air quality simulation approach to assess the accuracy of emission inventories for these applications, with the goal of identifying sources that are over or underestimated. This approach consists of finding constants of the linear combination of the estimated emission that maximize R2 and make the slope equal to one in the linear correlation analysis when the results from the air quality model are compared to the experimental measurements of air quality. This methodology was applied to the case of the mining region in northern Colombia. As one of the largest open pit coal mining regions in the world, this region consists of seven independent mines with no relevant additional sources of emission. Use of the proposed methodology allowed quantification of the amount by which companies over or underestimated their emission, as well as quantification of uncertainties due to sources not considered in the model but that locally affect each monitoring station.