<p>The presence of low-permeability interlayers hinders the migration and accumulation of the injected CO₂. The strong vertical and horizontal heterogeneity might complicate the accurate assessment of the CO₂ storage capacity within saline aquifers. In this study, a Sequential Indicator Simulation (SIS) constrained by horizontal probability trends was used to construct a sandstone distribution model for three different layers within the fifth member of the Permian Shiqianfeng Formation (P₃sh₅) in the Yulin area of the Ordos Basin. A porosity model was developed using a Sequential Gaussian Simulation. The relative error between sandstone distribution and geological understanding in the constrained model is controlled within 10%, so with a high matching degree. Previous studies often used averaging parameters among broad ranges that did not adequately account for reservoir heterogeneity. Consequently, the calculated storage potentials tend to be overly optimistic, potentially overestimating the actual capacity. Our study optimized the formula from the Carbon Sequestration Leadership Forum (CSLF), the grids were only used within the model characterised by porosity (Ф) &gt; 5% and thickness (h) &gt; 0.5 m instead of using the real average data. Three types of foreground areas are divided according to the different grades of porosity, thickness and saturation. Heterogeneity plays a role in determining the distribution of favourable storage zones. Favourable zones become more dispersed as heterogeneity increases. When compared to the storage capacity of two methods, the optimized one is compatible with heterogeneity, demonstrating smaller results. In summary, the proposed methodology improves the accuracy of both the geological model and the assessment of CO₂ storage capacity, thus providing an effective geological basis for subsequent storage site selection.</p>