Spatial Transferability of Expansion Factors for Estimating Pedestrian Volume at Intersections

Pedestrian exposure at intersections is a key input for jurisdictions to develop pedestrian-centric strategies. Exposure is typically expressed as the annual average daily pedestrian traffic (AADPT). The expansion factor method is the most common technique employed to estimate AADPT, consisting of the expansion of short-term counts (STCs) using factors calculated from sites with continuous counts (CCs). However, jurisdictions often lack sufficient CCs to determine expansion factors, which motivated this work of assessing the spatial transferability of expansion factors. This paper examines the spatial transferability across three jurisdictions in Ontario, Canada, with similar school holiday and weather seasonality, and one jurisdiction in Arizona, U.S.A., with different characteristics. It was assumed that the "base jurisdiction" has CCs available, while the "target jurisdiction" has only STC sites, and applies expansion factors from the base jurisdiction. Two approaches were tested: the single factor method, averaging expansion factors across all sites in the base jurisdiction, and modeling methods, which assign STC sites to factor groups based on models developed in the base jurisdiction. The single factor method showed acceptable transferability within the Ontario jurisdictions, with an average absolute increase of 2.3% (8.7% relative increase) in the mean absolute percent error of AADPT achieved when comparing the spatially transferred factors to those obtained when the base and target jurisdictions are the same. However, poor transferability was observed between jurisdictions with differing characteristics. The application of modeling methods yielded inconsistent results, possibly because of the limited number of sites available in this study, and requires further investigation.