Estimation of Mobility Flows at Sub-regional Level: An Application to Piedmont Based on a Socioeconomic Scenario

Abstract

Notwithstanding considerable efforts are being made to gather transport and mobility data, by means of new technological devices, in Italy a major spatially comprehensive source of information is the population census. Data about journeys-to-work and journeys-to-school at municipality level have been made available by the Italian Central Bureau of Statistics (ISTAT) at 1991, 2001 and 2011 census years. Such data provide a sound information basis for drawing retrospective accounts of mobility. However, it is likely to be unsatisfactory for a number of planning activities, such as monitoring and forward looking investigations. They often require expensive individual surveys which may be inaccessible, either for public and private agencies. This study aims at alleviating the problem by developing a computational and analytic tool for estimating journeys-to-work at sub-regional level. The study proposes a strategy which links the reconstruction of yearly mobility flows, based on available spatially fine-grained socioeconomic information, with the generation of new flow matrices, depending on regional socioeconomic scenarios. In this application to Piedmont, we first reconstruct the mobility flows in the 2001–2013 period according to sub regional population and employment data as well as the census mobility tables. Econometric regression techniques have been used to estimate the origin and destination totals of mobility matrices; Wilson’s entropy maximization approach to fully constrained spatial interaction models has been applied to compute the matrix cell values. The mobility deterrence parameter series associated with the flow matrices have been analyzed according to a set of socioeconomic variables, for which regional level demographic and economic trends from 2014 to 2020 are provided by ISTAT (population forecasts) and Prometeia (macro-economic studies). Parameters estimates have been obtained which, together with auto-regressively calculated origin and destinations totals of mobility tables, have been used to infer the regional flow matrices from 2014 to 2020. In discussing the main results of the approach attention focuses on its practical relevance for planning purpose and its portability. Some methodological issues concerning the calculation of the deterrence parameter are also mentioned.