The invention relates to an urban area function attribute identifying method based on a satellite image and a network, which belongs to the technical field of multi-mode sensor data processing. The method comprises the steps that the connection behavior information of users who are active on routers distributed in an urban area and the satellite image corresponding to the whole urban area are collected; based on the collected data, feature definition and extraction of the urban area are carried out, and specifically the behavior feature of users is extracted from network data, and a visual feature is extracted from the satellite image feature through a convolution neural network; and finally, an AdaBoost algorithm is used to integrate a series of decision tree models to identify the function attribute of the urban area, which increases the stability of a found key feature during training iteration. The method provided by the invention has the advantages that sensor data of two kinds of modes are merged; the flexibility and accuracy of urban area function attribute identifying are improved; and the method has the advantages of accessible data, low computational complexity of the algorithm and high scalability, and is suitable for practical industrial applications.