Abstract: Since 2010, coal production in China has increasingly shifted to the western regions, characterized by superior resource endowments and favorable mining conditions. The three major industries—coal, electricity, and chemicals—in the northwestern region have experienced significant growth, leading to substantial emissions of air pollutants and CO₂ during both production and operations. However, the limited number of existing ground monitoring stations hinders a comprehensive and precise reflection of the emissions induced by these industrial bases. In contrast, satellite remote sensing offers broader coverage at relatively lower costs. An atmospheric mass-conserving approach and the Tropospheric Monitoring Instrument (TROPOMI) onboard Sentinel-5 Precursor datasets are employed to quantify the emissions of NO_x, CO, and SO₂ from the northwestern coal-based regions from 2019 to 2020. Given the “co-synchronous, co-located, and co-sourced” characteristics of pollutants and greenhouse gases, CO₂ emissions are indirectly estimated based on established bottom-up emission relationships. The results show that: The physical and chemical driving factors derived from the mass-conserving equation appropriately reflect coal consumption levels, topographic factors, and emission characteristics in the northwestern regions, exhibiting consistency in annual distribution. An underestimation of coal-fired source pollutant emissions is also revealed in the northwestern regions according to existing inventories. Contributing factors include numerous near-zero value areas in the inventory despite prevalent residential coal burning and coal fires. Additionally, a lack of statistical data results in many emerging small and medium-sized coal-fired industries and other emission sources being inadequately recorded. The use of a single species for indirectly estimating CO₂ emissions leads to significant errors; Hence, integrating results from NO_x, CO, and SO₂ provides more precise estimates. The temporal and spatial distribution consistency is demonstrated with the Multi-resolution Emission Inventory model for Climate and air pollution research (MEIC). However, in 87% of valid grids, the CO₂ emission estimates exceed those reported by MEIC, underscoring the necessity of improving the accuracy of CO₂ emission inventory for both low and high emission scenarios.