Traffic flow models are important to describe various complex trafficprocedures on highway or road network and to establish intelligence trafficsystems (ITS). Lots of improvement has been made in traditional traffic flowmodels, and many new traffic flow theories have been developed.Generally the formation of a traffic flow model requires massive observationas well as in-depth analysis of real road traffic. Empirical data areindispensable to identifying parameters when the corresponding model isutilized. One of the basic demands of collecting such data is repeatability,that is, a large amount of measuring must be conducted under nearlyidentical circumstances to provide the data and model parameters.It is drawn in this paper that over 26000 ``car velocity-headway distance''data pairs are obtained from the long time video recordings of a section inYanan Expressway of Shanghai with three traffic circumstances: peak hour;low traffic hour; snow day. This way to collect traffic data is differentfrom some traditional methods in which sensors are fixed along the road orburied underground to measure the ``car velocity-traffic flow'' datapairs. The obtained ``velocity-headway distance'' data pairshave better synchronized characteristics and are easy to transform into``velocity-density'' or ``flow-density'' data pairs for further research.After comparison study of manifold well-known traffic flow velocity-densitymodels based on above measuring data, the superiorities of one-dimensionalpipe-flow model are found in calculating thetraffic flow parameters of expressway in our country.Since the key parameter in 1-D pipe-flow model, named traffic behaviorparameter m, can be adjusted, a basic method is proposed to draw thenon-linear characteristics of traffic flow by 1-D pipe-flow model withchanging parameters.