In the era of big data, traffic flows play an important role in understanding our socioeconomic environment. In recent years, two types of traffic flows, smart card transactions and taxi GPS trajectories, have been widely but usually separately utilized to understand human mobility in big cities. To date, although numerous research achievements have been made, the relationship between these two types of traffic flows that occur in the same area and in the same time period is still poorly understood. Thus, the pattern of urban human mobility may be biased by using a single type of traffic flow. In this study, we aimed to compare the urban human mobility patterns derived from the two traffic flows, i.e., smart card transactions and taxi GPS trajectories. Taking the area within the Sixth Ring Road of Beijing as study area, we collected the smart card transactions and taxi GPS trajectories data from May 9^th to May 15^th in 2016. Specifically, we compared: ① the spatio-temporal distributions of public transit and taxi usage; ② the travel distance of public transition and taxi usage and the distance-decay effects, and ③ the spatial community structures discovered from the two traffic flows. Our results show that: ① the spatial distributions of travel demand revealed by two traffic flows are highly correlated. However, the correlations between origin and destination time series extracted from the two traffic flows were very weak; ② the usages of public transit and taxi had spatial heterogeneity, and the spatial difference between the usages of public transit and taxi cannot be fully explained by the number of intersections, the number of points of interest, and the average distance to the nearest city center, shopping center, hospital and subway; ③ the travel distances extracted from taxi GPS trajectories decayed faster than that extracted from public transit, indicating that public transit was more important in facilitating long-distance travel, and ④ spatial communities discovered from the two traffic flows both reflected the polycentric spatial structure of the city. However, the differences in spatial community structures indicated that the two traffic flows played different roles in spatial interactions in the city. The quantitative comparison between smart card transactions and taxi GPS trajectories could improve our understanding of human mobility in Beijing, which also demonstrates the potential biases by using a single traffic flow to study urban system dynamic. Our results suggest integrating multi-source traffic flows to understand urban human mobility patterns in future.