Abstract: Understanding the impact of open-pit mining activities in desertified areas on soil respiration (Rs) is crucial for assessing the carbon cycle in the fragile ecosystems of open-pit mining areas. Current studies on Rs in mining areas mostly rely on field sampling methods, which are difficult to achieve large-scale and multi-temporal Rs remote sensing estimation. The Hongshaquan open-pit mine and its surrounding areas are taken as the research object, with Rs estimation results from unmanned aerial vehicles (UAVs) used as the sample data for modeling. Six methods—including multiple linear regression (MLR), random forest (RF), back propagation neural network (BPNN), support vector machine (SVM), particle swarm optimization-based support vector machine (PSO-SVM), and stacked auto encoder (SAE)—are employed to construct the optimal Rs estimation model. This model aims to estimate the spatial distribution of Rs in the mining area and analyze the changes in Rs intensity and range with different mining years. The results show that: The Rs estimation model based on RF has the highest accuracy, with a coefficient of determination (R²), root mean square error (RMSE), and Akaike's information criterion (AIC) of 0.762, 0.514 μmol/(m²·s), and 5714.1, respectively, in the validation set. The reliability of the estimation results in different years was verified using the soil organic carbon index (SOCI). The results show that Rs in the three years has a significant positive correlation with SOCI, indicating that the UAV estimation data can be used as sample data for satellite modeling to achieve satellite-scale Rs estimation. The soil stripping and coal dust deposition caused by coal mining activities significantly increase the Rs intensity in the mining area. The mean Rs in the mining pit and its surrounding areas reaches 2.71 μmol/(m²·s), which is about 5% higher than that in the unmined areas. With the increase in mining years, the Rs value of the bare land around the mining area increases from 2.35 μmol/(m²·s) in 2019 to 2.55 μmol/(m²·s) in 2023, with an increase of about 10%. With the increase in mining years, the influence range of the mining pit on Rs increases from less than 3 km to more than 3 km, indicating that the impact of mining activities on Rs is not only limited to the surrounding areas of the mining pit but also extends to a wider area, with an influence distance of approximately 3 km. In conclusion, the UAV estimation results can be used as sample data for satellite-scale Rs estimation. Combined with satellite images of different years, it can achieve Rs remote sensing estimation in the mining area, clarify the variation pattern of Rs, and determine the influence range of the mining pit on Rs under different mining years, providing a scientific basis for carbon emission monitoring and management in the mining area and even wider regions.