Spatial prediction of AADT in unmeasured locations by universal kriging and microsimulation of vehicle holdings and car-market pricing dynamics

Chapters 1 through 5 of this thesis explore the application of kriging and geographically weighted regression (GWR) methods for prediction of average daily traffic counts across the Texas network. Accurate measurements of traffic are essential for proper planning and management of pavements, roadway upgrades, congestion mitigation, and other aspects of ground-based transport. Results based on Euclidean distances are compared to those using network distances, and both allow for strategic spatial interpolation of count values while controlling for each location’s roadway functional classification, lane count, speed limit, employment density, and population access. Both universal kriging and GWR are found to reduce errors (in practically and statistically significant ways) over non-spatial regression techniques, though errors remain quite high at some sites, particularly those with low counts and/or in less measurement-dense areas. Nearly all tests indicated that the predictive capabilities of kriging exceed those of GWR by average absolute errors of 3 to 8 percent. Interestingly, the estimation of kriging parameters by network distances showed no enhanced performance over that with Euclidean distances, which require less data and are much more easily computed. Chapters 6 through 10 explore vehicle purchase and use decisions, which can be central to estimates of crash outcomes, emissions, gas-tax revenues, and national energy security. An auction-style microsimulation of fluctuating vehicle prices is combined with a random-utility-maximizing choice model to produce a model for the evolution of personal-vehicle fleets, recognizing both used- and new-vehicle markets. All buyers and available vehicles are enter the auction process for vehicle selection, with demand, supply and price signals of used cars endogenous to the model. The thesis describes the modeling framework in detail, along with its implementation using Austin, Texas data (for behavioral parameters and a synthetic population). The fleet dynamics are simulated over a 20-year period, highlighting the model’s flexibility and reasonable response to multiple inputs and contextual scenarios. A simulation of doubled gas prices showed a large increase (10%) in the share of the sub-compacts, with smaller decreases in pickup trucks, vans and large cars. A high scrappage rate, sometimes employed to increase turnover, resulted in used-vehicle sales falling by 12%, and new-vehicle sales growing by 3%.