Estimating Latent Travel Demand from Open Data: Validation and Regional Variation in Demand Realization

Transportation planning in regions with insufficient public transit often relies on observed travel volumes, overlooking unmanifested latent demand. This study proposes a method to estimate latent travel demand (LTD) within a few kilometers of residential areas using open data—population distribution, aggregated activity data, and facility locations—and investigates its validity and regional variation across three Japanese prefectures (Hiroshima, Ibaraki, and Iwate). Validity was assessed by examining the relationship between estimated LTD and observed travel volumes (apparent traffic volume, ATV) derived from smartphone GPS data. Three key findings emerged. First, LTD, combined with facility count, transit availability, and distance, explained observed travel volumes with R2 = 0.61–0.63 on the original scale and R2 > 0.91 on the log–log scale, and likelihood ratio tests confirmed that LTD provides information complementary to, and distinct from, conventional accessibility indicators. Second, regional fixed-effects analysis revealed that while LTD is a significant predictor across all regions, the demand realization rate (ATV/LTD) varies systematically—from 0.021 in transit-rich Hiroshima to 0.045 in car-dependent Iwate—reflecting differences in transit infrastructure and car dependency. The interaction between LTD and transit availability was statistically significant (p = 0.008), with moderate-transit areas showing the largest marginal effect of LTD on realized travel. Third, activity-type analysis showed that mandatory activities (e.g., medical visits, childcare) maintain stable demand realization regardless of transit levels, while discretionary activities (e.g., hobbies, leisure) are strongly transit-dependent. These results demonstrate that LTD estimation offers a practical, scalable diagnostic tool for identifying mobility gaps and prioritizing transit investments in resource-constrained municipalities.