New data sources for Adelaide’s transport demand modelling
In the recently completed project Recalibrating Adelaide’s strategic transport model, the Department of Infrastructure and Transport (DTI) and the University of South Australia’s Institute for Choice looked not only to update the data sources informing its transport demand modelling and infrastructure development, but to prove the methodology to be efficient and cost-effective. The final report for the project is downloadable below.
Methodology
The Metropolitan Adelaide Strategic Transport Evaluation Model (MASTEM) comprises ten sub-models that are used together to model and forecast travel within the Greater Adelaide (GA) metropolitan region, as shown below in Figure 1.
Four of the sub-models (household stratification, car ownership, network assignment and external traffic) were recalibrated using Greater Adelaide (GA) data collected by other organisations, such as the Australian Bureau of Statistics and Tourism Research Australia.
For five of the travel-related sub-models (trip production, trip attraction, travel market segmentation, trip distribution and mode choice), household travel surveys (HTS) were used, and transport cost skims from Melbourne, Southeast Queensland and Perth in Australia and Auckland from New Zealand to estimate model parameters for each of these jurisdictions.
Additional 1-day travel diary data was collected through a web-based survey from 493 residents in GA in late 2021. This data helped identify appropriate parameters for GA from the set of parameter estimates for the other jurisdictions. The freight sub-model was ultimately excluded from the scope of the present study, due to unavailability of appropriate freight movement data.
Results
The ABS and other organisations already collect datasets that offer useful information for strategic transport demand modelling. For example, the Census collects car ownership information from all households every 5 years.
In comparison, a HTS would only collect this information from a small fraction of households, typically less than 1 per cent of the total population, often at less frequent intervals.
These other datasets have previously not been used for the calibration of MASTEM. For the household stratification and car ownership sub-models, it was found that these datasets were more reliable than a traditional HTS. However, in the case of the external traffic sub-model, there are significant gaps in these datasets, and additional data might need to be collected to produce credible results.
HTS data from other jurisdictions offers useful insights on travel patterns within GA. In general data from the three Australian cities was in close correspondence with each other, and agreed well with available information about GA. Auckland was an outlier in many cases, having smaller household sizes, lower rates of car ownership, and higher shares for single-occupancy car trips.
By including multiple jurisdictions within the analysis from which to pick parameter estimates for the GA region, outliers were more easily identified. By collecting primary travel diary data from a small sample of GA residents jurisdictions were identified that most closely corresponded to GA.
In conclusion eight of the ten sub-models within MASTEM were recalibrated to a high degree of confidence, using available data from other organisations and other jurisdictions, augmented only by a small-scale primary travel diary data collection within the GA region.
Key benefits
The methodology developed by this study offers a cheap and quick alternative for the calibration of strategic travel demand models, when compared to existing methods based on primary HTS data collection.
Researchers were able to develop and implement the methodology in less than one year at a cost of roughly $250,000. In contrast, a primary HTS data collection exercise in the GA region alone would cost more than $1 million, on top of which would be additional time and money costs associated with model estimation.
What’s next?
The recalibrated MASTEM parameters are currently being integrated within the new Strategic Adelaide Model (SAM). We are sourcing additional HTS and transport cost skims data from other jurisdictions, such as Hobart and the Australian Capital Territory, to increase the robustness of our analysis.
Finally, other sources for freight movement data are being explored. Subsequent research could examine how HTS data from different jurisdictions could be combined and reweighted to create a ‘pseudo-HTS’ dataset that is representative of GA, in terms of both the demographic characteristics of the local population, and the level-of-service of the local transport system.
This pseudo-HTS can then be used directly to estimate appropriate parameters for GA. The proposed approach can be validated using findings from the present study and, if successful, can subsequently be applied to the calibration of analogous strategic travel demand models for regional SA.
What’s next?
Download your copy of the final report, Recalibrating MASTEM using data from the ABS and other jurisdictions, by clicking the button below.
Discover more from iMOVE Australia Cooperative Research Centre | Transport R&D
Subscribe to get the latest posts sent to your email.