Publication Detail

Exploratory Analysis: A Method to Improve the Link Between Travel Demand Model and DTIM2


Research Report

Suggested Citation:
Lin, Kuo-Shian and Debbie A. Niemeier (1998) Exploratory Analysis: A Method to Improve the Link Between Travel Demand Model and DTIM2. Institute of Transportation Studies, University of California, Davis, Research Report UCD-ITS-RR-98-06

This study specifically focuses on the relationship between the travel demand model and the Direct Travel Impact Model (DTIM2) (CalTrans, 1994), and seeks to improve the estimation of gridded, hourly emission inventories using TDM simulation outputs.

The objectives of this research are to:
  • Explicitly define the links between TDM and air emissions models as evidenced by inputs and outputs exchanged between the models;
  • Develop a statistical model expressing the relationship between hourly traffic counts and TDM simulation volumes by trip propose;
  • Propose and test a method for allocating the "time-period-based" TDM assignments into "hourly breakdowns by trip purpose".
The study begins with a review and assessment of the broad family of transportation-related emission estimation models, then explicitly defines the links and weakness of those links between the existing TDM and DTIM2. A statistical model is then developed based on TDM simulation and traffic count data.

Throughout the modeling and verification processes, the model variables will be discussed and the hourly by trip purpose "allocation factors" for the TDM network will be defined. According to model interpretation, the "allocation factors" can be used to transfer the conventional three "time-period" (AM, PM and Off-Peak) TDM simulations into 24 "hourly" resolutions, and therefore can serve as an important input for DTIM2. Thus, the regional gridded, hourly emission inventories estimated from DTIM2 should become more persuasive, and this in turn assures reasonable estimation of emissions input for the EPA's DAM photochemical air quality model.

Study Contributions

This study contributes to the improvement of air quality forecasting by proposing an exploratory statistical model for interfacing TDM, DTIM2, and the photochemical air quality model, DAM. A second contribution comes in verifying the applicability of a set of statistical models to the analysis of the TDM validation problem, and the expansion of the statistical models to the improvement of gridded, hourly emissions estimation. An improvement in the emissions estimation of EPA's air quality models, DAM, can be expected by using the models proposed in this study to enhance the weak link between TDM and DTIM2.

A third contribution is that the study defines the type of research that is needed to facilitate testing and use of the exploratory models. This research proposes procedures for handling missing data; we propose an application for replacing missing data using Kalman estimators from state-space modeling to fill missing hourly counts for time series data. Finally, this study also contributes to the long-term objective of developing methods that improve the development of the Statewide Implementation Plan.

Report Organization

The study is divided into five chapters. Chapter 2 conducts a broad review of transportation related emissions models and highlights the DTIM2 and TDM related critical interaction issues. The focus is on the relationship between the models, the state of the practice, possible interface models, and the problems associated with the current state of the practice.

Chapter 3 presents the proposed modeling technique for improving the link between TDM and DTIM2 using a set of Multivariate Multiple Regression (MMR) models. The conceptual MMR models can be used to validate the TDM simulation on a disaggregate basis (i.e., trip assignments stratified by trip purposes). The enhancement for the link between the TDM and DTIM2 is expected to be achieved by extending the application of the MMR model by using the model coefficients as "allocation factors" to allocate the time-period-based TDM assignments into hourly breakdowns. The chapter concludes by describing proposed additional research using Monte-Carlo simulations to verify the proposed conceptual allocation models.

Chapter 4 describes the empirical estimation of the proposed conceptual model. Using Sacramento data, the preliminary study of the allocation model is presented and preliminary findings are discussed. The chapter also discusses the limitations and study issues for further comprehensive empirical study of the dissertation. Finally, Chapter 5 briefly reviews the additional research needed and develops a proposed lists of tasks to extend and verify the model results.
Ph.D. Dissertation