Sample home interview data from travel surveys undertaken in reading in 1962 and 1971 were used to test the ability of several simple trip distribution models to forecast the distribution of internal trips within the survey area over the nine year interval. The behaviour of the Furness growth factor model and a simple exponential gravity model was investigated in some detail, while a simple proportions model and a uniform factor model were considered in less depth. Two levels of zonal representation were used and several trip categories were considered. The Furness model performed better than the other models and, for a given flow size, better forecasts were obtained for the home-to-work trip category than for all purposes together. The root-mean-square error in the forecasts from the different models varied from 40 to 250 per cent of the mean flow size but a large part of the error could be attributed to sampling errors: as might be expected from sampling theory, larger flows gave smaller percentage errors. The forecasting errors of the Furness model were associated with poor distribution of short trips and the over-estimation of intra-zonal flows, especially near the town centre. The relationship between forecast error and flow size was similar to that found in studies in the USA. (Author/publisher)
Samenvatting