A great deal of time and money is spent by local authorities on measuring traffic flows. In major surveys labour is often recruited to count and classify vehicles for one or more days at each of several sites in the study area. However, where the pattern of traffic flow variation through the hours of the day is roughly similar for all the sites in a group, one may be able to omit the counts during some hours at some of the sites, and estimate the 'missing' flows as proportional to those measured during the corresponding hours at the remaining sites. Carrying this idea to its logical conclusion, this paper investigates the design and use of sampling schemes in which the flow at each site, in a group of sites believed to have similar characteristics, is measured in a selected number of the sampling intervals into which the survey period is arbitrarily divided. The data are then pooled to give estimates both of the flows at each site in each interval, and of the total flows for the survey period. Such an approach can be relatively more efficient than conventional methods, especially where labour resources are limited, and an example is given of its application to a small town in the u.k. the basic principles apply also to the related problem of estimating the average daily total flow at a single site, from one-hour flow samples taken at intervals over a period of several days, weeks, or months. In this way, sampling fluctuations caused by abnormal events or weather conditions can be minimised, and trends and seasonal variations allowed for. The theoretical approach is given. For the covering abstract of the symposium, please see irrd abstract no. 224453.
Abstract