Access to urban parks is essential for promoting public health, social equity, and sustainability in cities. Most park planning guidelines only provide a pre-defined service area, failing to reflect the actual service area. Although a few recent studies have used mobile signal data to calculate the origin-park distance to identify park service areas, the limitations in data and methods could yield biased results, and an investigation of car-based park users is missing. Therefore, this study proposes a data-driven framework that leverages emerging connected-vehicle data to identify park service areas specifically for car-based users. The connected-vehicle data is first processed to identify trips between users' origin locations and parks, and their travel distances to parks are extracted. Furthermore, the mixture models, Gaussian Mixture Model (GMM) and Lognormal Mixture Model (LMM), are applied to fit the travel distance data and determine the optimal models. The evaluation results show that the LMM outperforms other models, and two distributions are sufficient to accurately fit the travel distance distribution. The proposed methods were implemented in Pima County as a case study to demonstrate their applicability. The results show that most municipal parks have service areas ranging from 8 km to 20 km, while county parks have slightly larger service areas. Furthermore, three federal parks have service areas ranging from 20 km to 30 km. All urban parks can be clustered into two groups using the K-Prototypes clustering method. The clustering results show that parks with smaller areas and a diversity of athletic facilities tend to have relatively small service areas, while parks with larger sizes and outdoor recreational facilities tend to have relatively large service areas.
扫码关注我们
求助内容:
应助结果提醒方式:
