Due to the growth of air traffic, some major airports in Japan may expect surface traffic congestion, which does not only cause increase of taxiing time and consequent fuel consumption, but also affects punctuality and reliability of flight scheduling. To cope with this issue, we have to develop some kind of decision support systems or operational procedures to keep airport operation smooth and punctual. The core technology of such systems and procedures is the prediction of surface movement time. With the aim of developing such kind of predicting function, we have tried an empirical analysis of surface traffic flow at Haneda Airport to extract the characteristics of congestion during surface movement, based on multilateration surveillance data. Focusing on departure traffic flow, we found some factors which determine and influence the surface movement time. In this paper, we describe the characteristics of elapsed time at other ground operation phases before lining to the queue, and also describe the characteristics of waiting queue growth at departure runway when the airport is busy. Finally, we discuss the capability of modeling departures' surface movement time based on these characterization, for predicting and managing surface movement time.