Abstract: To address the technical challenge of accurately reconstructing the gas content distribution field in mines under scenarios with limited computational resources, a coupling algorithm based on Universal Kriging (UK) and the Empirical Variogram Function (EVF), namely UK-EVF, was proposed to achieve high-precision spatial distribution characterization of gas content with lightweight computation. First, gas content data at different locations in the target area were collected through field measurements, and a dataset including spatial coordinates of measurement points and corresponding gas content was constructed. Then, the UK algorithm was introduced, and spatial parameters obtained by EVF were used as constraints. The EVF model was employed to quantitatively analyze the spatial structure characteristics of gas content distribution, and key spatial correlation parameters such as range, nugget effect, and sill were accurately extracted. On this basis, spatial interpolation was performed to predict gas content in unsampled regions, thereby achieving the reconstruction of the gas content distribution field in the prediction area. Four classical spatial interpolation algorithms and five mainstream machine learning and neural network models were selected for comparative analysis. The results showed that the UK-EVF coupling algorithm performed best among all compared methods. The absolute value of the mean relative error of gas content prediction was consistently controlled within 5.1%, and the runtime for distribution field reconstruction did not exceed 10 s. Under small-sample and low-computational-power scenarios, high-precision reconstruction of the gas content distribution field was achieved.