Large-scale vertiport network design for Innovative Air Mobility (IAM) faces computational scalability challenges, as regional candidate generation can yield tens of thousands of feasible parcels and millions of binary variables in Mixed-Integer Linear Programming (MILP) models. Yet existing methods lack systematic pre-optimization filtering to reduce the decision space while preserving operational and geographic realism. This study proposes a terrain-aware, data-driven spatial reduction strategy that serves as a pre-optimization candidate screening stage and investigates associated computational challenges in regional-scale vertiport network design. By integrating terrain-based distance accumulation with Spatial Principal Component Analysis (sPCA), correlated accessibility and land-use criteria are synthesized into structurally consistent decision-space representations. sPCA mitigates multicollinearity among spatial determinants and extracts dominant structural drivers without subjective weighting. Applied to a regional case study in Saxony, Germany, the framework reduces 74,944 feasible grid cells to 236 candidate locations, decreasing MILP binary variables by two orders of magnitude and reducing solution times from more than 96 hours to 18 minutes under identical solver configurations. The proposed approach enables integration of GIS-based spatial analytics and operations research, supporting geographically consistent and computationally tractable vertiport network planning for large-scale IAM systems.