Nitrogen oxides (NO<sub>x</sub> = NO + NO<sub>2</sub>) are key pollutants that contribute to ozone and secondary aerosol formation, posing environmental and health risks. Accurate simulation and forecasting of NO<sub>x</sub> pollution is essential for developing mitigation strategies. Local inventories in Thailand are infrequently updated, leading researchers to use global inventories such as CAMS-GLOB-ANT for simulation. Global inventories carry uncertainties due to assumptions in emission factors, outdated activity data, and coarse temporal resolution. To address these limitations, this study applies a top-down approach to update NO<sub>x</sub> emissions in Thailand using the iterative finite difference mass balance (IFDMB) method. Tropospheric NO<sub>2</sub> vertical column densities (VCDs) from the GEMS are integrated with the WRF-Chem to refine CAMS-GLOB-ANT emissions for September 2023. The simulations with posterior emissions are evaluated against TROPOMI NO<sub>2</sub> VCDs and surface NO<sub>x</sub> concentration. Results show that the baseline simulation overestimates NO<sub>2</sub> VCDs across Thailand compared with GEMS, except in North Thailand. Consequently, IFDMB reduces NO<sub>x</sub> emissions across most regions but increases in the North. These adjustments improve model bias and error relative to GEMS. However, when evaluated against TROPOMI, we find an increase in the bias for North Thailand, likely due to discrepancies between GEMS and TROPOMI retrievals. Discrepancies between GEMS and TROPOMI highlight the importance of future calibration across satellite products. Comparisons to surface observations indicate that IFDMB shifts the NO<sub>x</sub> peak to later than observations. This is because observations are strongly influenced by local transportation sources, which are hard to observe and simulate by GEMS and the model, respectively.