Annual Average Daily Traffic (AADT) is used in several planning, roadway design, operational and safety analyses by transportation planners and engineers. Existing methods are very complex and do not adequately address the modeling needs. Errors and inaccuracies in a traditional four-step method get carried to later steps often resulting in incorrect estimates of travel demand. The primary focus of this research is to develop a systematic and simplified methodology to estimate link level travel on roadways. The proposed methodology involves scientific principles and statistical techniques, but bypasses the tedious four-step method. Two spatial methods, first one based on "spatial proximity" and second one based on "spatial weighting", are proposed to estimate link level travel. While the former method investigates to identify ideal "proximal" distance to capture spatial data, the later method involves application of "spatial weights" that decrease with an increase in distance to integrate spatial data from multiple buffer bandwidths. Generalized Estimating Equations (GEE) models are developed for both the methods using Poisson and Negative Binomial distributions with and without network characteristics to facilitate transportation planning and analysis. Validation of the developed models is carried out using Chi-Square Statistic test. The goodness of fit statistics indicates that Negative Binomial models performed better than Poisson models. Models with network characteristics performed better than models without network characteristics. Model validation results indicate that link level travel can be accurately estimated using both the spatial methods.
