Calibration and Validation of Large Models- The Challenges Facing Highways England’s Regional Traffic Models
Dr Paul Read, CH2M, Graham Powell, Highways England, Chris Bruce, Arup
The paper outlines the challenges in calibrating large models developed using big data sources. It will refer to the lessons learnt, collaboration and innovations in the 5 new English regional models which will support future transport investment.
The increased availability of big data allows the development of larger transport models to support robust decision making to develop infrastructure programmes. However, this leads to increased challenges in calibrating and validating such large scale applications to ensure the models are fit for purpose over such a large area. In scheme development programmes the production of a traffic model is often seen as a major potential risk to time and budget. With new sources of ‘big data’ and data fusion techniques there are opportunities to understand travel patterns, but models still require detailed calibration and validation within reasonable time frames.
Highways England commissioned five Regional Traffic Models (RTMs) covering the whole of England in order to provide consistent evidence bases to develop schemes at a regional level. These are among the largest spatially detailed models developed in England but still require demonstration that they are fit for purpose. Given that five different study teams, across a number of companies, are producing these models, the approach adopted has been one of collaboration, to a level unprecedented in the industry.
The challenge has been to provide guidance to enable robust calibration and validation that meets the strategic and local needs for these models. In addition, we are working with a wide range of data sets from mobile phone matrices, networks developed from ITN and a wide variety of traffic count and journey time data. Therefore innovative approaches have been required to enable the study teams to both understand the complexities of the input data and the outputs from the models. Highways England established Technical Consistency Groups to ensure best practice was shared across the RTMS. One of these related to Calibration and Validation of the RTMS and this group developed relevant guidance on the approach and the techniques adopted.
This paper outlines the calibration and validation issues tackled in the production of the RTMs. The paper will provide the background to the RTMs and the need for consistency of approach. We will describe the approaches adopted for calibration and validation and draw on the guidance that was developed to assist the production of the RTM. The paper will explain how we used the wide ranging ‘big data ‘ sets to really understand the travel patterns in the RTM areas but also to provide a rich source of independent data to assess the quality of the models. Our challenge was that the focus was not just on a single defined area for a scheme, but potentially anywhere across the Strategic Road Network. The paper will show how the group developed procedures to assess model quality across not only wider areas but also at individual link level given the need for more robust information to be used for air quality assessment. We will show the phased approach adopted to calibration and validation and highlight the lessons learnt in producing the RTMs. Examples will be given from each of the RTMs to demonstrate the robustness of the resulting models which provide a legacy for scheme assessment over a number of years.
Association for European Transport