Analyzing the spatio-temporal variations of atmospheric carbon concentrations in energy resources development areas and identifying influencing factors are crucial for exploring a high-quality development pathway in the context of “Dual Carbon”. Xinjiang Uygur Autonomous Region serves as a vital base for energy and strategic resources in China. This study oriented to the current status of energy resource development in Xinjiang Uygur Autonomous Region, we collected and preprocessed Orbiting Carbon Observatory-2 (OCO-2) carbon dioxide Level 3 data products from 2015 to 2021. We analyzed the temporal trends and spatial distribution patterns of atmospheric carbon concentration in the study area and structured a deep forest regression model to analyze the driving factors of the spatio-temporal variations in carbon concentration. The results indicate that: (1) Xinjiang Uygur Autonomous Region, Junggar Basin, Turpan-Hami Basin and Tarim Basin’s XCO₂ concentration exhibited a cyclic upward trend from 2015 to 2021, with a “decrease-first then increase” growth rate, showing a distinct “high in spring, low in winter” seasonal variation trend; (2) in spring, autumn, and winter, the spatial pattern of XCO₂ concentration in Xinjiang shows a “high in the north, low in the south” trend, with high XCO₂ concentrations accumulating in basin and energy resource development areas. Conversely, a trend of “low in the north, high in the south” observed in summer; (3) topographic relief, wind velocity, NDVI, land surface temperature, precipitation, 10-meter V wind, 10-meter U wind, and energy development intensity significantly influence the spatio-temporal distribution of regional XCO₂ concentration, showing notable spatial heterogeneity and significant differences. These findings contribute to understanding the mechanism of carbon concentration evolution in energy resource extraction areas and hold profound implications for achieving national carbon reduction targets, guiding carbon neutrality strategies, and monitoring carbon emission reduction effects.