Effect of Weather on Traffic Flow

IN order to estimate trends in road traffic, censuses were held, in conjunction with the Ministry of Transport, at twenty-one selected points over Great Britain during the weeks August 15–21, 1949, and August 14–20, 1950. In general, the weather was fine during the 1949 census week and wet during the 1950 one. A comparison of he traffic flows for the two Sundays (1949 and 1950) showed a very marked reduction in the flow for 1950, when, owing to the lifting of petrol restrictions, it would have been expected to show an increase. It seemed likely that this observed reduction was due to weather conditions, and examination of the traffic and weather data has shown that a high negative correlation exists between rainfall and traffic. For example, if we compare the total number of solo motor-cycles observed on each of the days Monday to Friday in the 1950 census with the daily rainfall (the average of the rainfall recorded at the meteorological stations nearest the census points), the coefficient of correlation is − 0.91 (significant at the 5 per cent level). For the two Sundays, we must use a different method; we compare the percentage change, between 1949 and 1950, in flow of solo motor-cycles at each of the twenty-one points with the amount by which the rainfall at each point in the week in 1950 exceeded that in 1949. This gives a correlation of − 0.72 (significant at the 1 per cent level). Results of calculations such as these are summarized in Table 1. As the two Saturdays were both fairly fine, no direct method was available for determining the weather effect; but for the purpose of Table 2 it has been assumed that it will lie between that for Monday to Friday and that for Sunday.