<p>As urban populations grow, climate-related risks in cities are expected to intensify. Urban climate simulations have advanced considerably, particularly in representing the heterogeneity of complex urban structures and the dynamical impacts of built environments through Urban Canopy Models (UCMs). However, these advances introduce new sources of uncertainty and require locally adapted parameters. This study focuses on the Metropolitan Area of Buenos Aires (AMBA), one of the largest urban conglomerates in Latin America, and examines three key components of urban climate simulations. First, a global urban classification dataset was refined using local data. Second, UCMs available in the WRF model are evaluated. Third, a sensitivity analysis is conducted on the most complex UCM to assess the impact of different urban canopy parameters (UCPs). Incorporating local building height data and shapefiles of representative areas (e.g., low-development neighborhoods) results in a more accurate urban surface map for AMBA. During a 10-day heatwave, results suggest that model performance depends more on UCM selection than on its internal configuration. The integration of multilayer UCMs, refined urban classification, and improved UCPs slightly enhances the agreement between modeled and observed surface variables at 10 urban meteorological stations. The sensitivity analysis of UCPs is performed using a signal-to-noise ratio approach in which noise is estimated from internal model variability. Surface temperature and wind speed are particularly sensitive to building height, urban fraction, air conditioning use, and street direction. These results highlight the importance of prioritizing these parameters to improve the reliability of urban climate simulations.</p> Graphical Abstract <p></p> <p>This graphical abstract illustrates the study area, the Metropolitan Area of Buenos Aires (AMBA), and three mail components analyzed for urban climate simulations. (1) Local Climate Zones (LCZs), (2) Urban Canopy Models (UCMs), and (3) Urban Canopy Parameters (UCPs). First, a reclassified LCZ map is shown, highlighting how the use of local data enhances spatial detail and increases the diversity of the urban environment, as seen in the distribution of LCZ percentages across the 10 urban classes. Second, different UCMs available in the Weather Research and Forecasting (WRF v4.5.1) model were tested. The surface temperature Taylor Plots for 10 weather stations and their mean demonstrate that the selection of UCM has greater influence on the model performance than its configuration, although revised parameters provided slight improvements in the representation of surface variables. Finally, UCP sensitivity was assessed through 19 simulations with fixed changes in the numerical value of certain UCPs. Sensitivity, quantified using the Signal-to-Noise ratio against internal variability, is presented both as numerical values (statistically evaluated) for each experiment and as maps for the whole AMBA region. Results indicate that surface variables, particularly temperature and wind speed, are most sensitive to morphological and air conditioning parameters.</p>