<p>Evaluation of water richness in sandstone is an important research topic in the prevention and control of mine water disasters, and the water richness in sandstone is closely related to its porosity. The reflection seismic exploration data have high-density spatial sampling information, which provides an important data basis for the prediction of sandstone porosity in coal seam roofs by using reflection seismic data. First, the basic principles of the variational mode decomposition (VMD) method and the random forest method are introduced. Then, the geological model of coal seam roof sandstone is constructed, seismic forward modeling is conducted, and random noise is added. The decomposition effects of the empirical mode decomposition (EMD) method and VMD method on noisy signals are compared and analyzed. The test results show that the first-order intrinsic mode functions (IMF1) and IMF2 decomposed by the VMD method contain the main effective components of seismic signals. A prediction process of sandstone porosity in coal seam roofs based on the combination of VMD and random forest method is proposed. The feasibility and effectiveness of the method are verified by trial calculation in the porosity prediction of model data. Taking the actual coalfield reflection seismic data as an example, the sandstone porosity of the 8 coal seam roof is predicted. The application results show the potential application value of the new porosity prediction method proposed in this study. This method has important theoretical guiding significance for evaluating water richness in coal seam roof sandstone and the prevention and control of mine water disasters.</p>