<p>Assessing the performance of an atmospheric dispersion model to estimate nitrogen dioxide (NO₂) remains challenging in regions with limited monitoring, where observations are scarce and key variables like ozone (O₃) are often unavailable. In this study, we explore different methodologies to assess the influence of chemistry on modelled high NO₂ concentrations using DAUMOD-GRS, an urban atmospheric dispersion model with a simplified chemical scheme, applied to the Metropolitan Area of Buenos Aires (MABA) over four years of hourly simulations. We evaluate three complementary approaches: (i) clustering of input/output conditions associated with NO₂ exceedances; (ii) analysis of the NO₂/NOx concentration ratio; and (iii) sensitivity simulations to key parameters in the chemical scheme, such as reaction time and background O₃. Clustering proved useful for detecting metadata issues and identifying consistent conditions associated with exceedances. Our results highlight that, although the NO₂/NOx ratio is widely used as a chemical diagnostic, its value depends strongly on the accuracy of NOx estimates, which can limit its applicability for evaluating chemical processes in practice. A bivariate analysis of modelled-to-observed concentration fractions of NO₂ vs NOx proved useful for quickly visualizing the distribution of errors in NO₂, NOx, and their ratio, along with associated input data conditions. Sensitivity tests revealed that the fraction of NO₂ in NOx emissions had greater impact on hourly exceedances, while background O₃ had a larger influence on annual means. These methods offer complementary insights and are particularly valuable in data-scarce contexts. Future work may benefit from incorporating O₃ observations and comparisons with chemical transport models to further support chemical evaluations and to explore additional reactions beyond those included in simplified schemes.</p>