Different from the traditional embedded system or the real-time system, Cyber-Physical System (CPS) integrates computation with physical entities, making it easy for people to interact with the world that we live. With the development of CPS, more devices will be integrated into the internet, especially in the field of Environment Monitoring. Thus, the management and remote control of devices cause issues for CPSs. Towards this end, we develop a CPS by proposing a scalable, flexible architecture in Environment monitoring. The environment monitoring system (EMS) consists of three layer: physical sensing layer, data transport layer, data management layer. Generally, sensors and sensor hub which collects sensor data and transfer them to the upper layer make up of physical sensing layer. And the main function for data transport layer is to forward sensor data to the specified destination. The last layer is in the cloud, composed of a data processor and a knowledge base. Data from physical sensing layer can be analyzed due to the strong computation ability of cloud computing. Also, there are abundant restful API (application programming interfaces) are provided by management layer based on users' demand. The proposed architecture provide convenient management with devices since all the information of devices are stored into database in the cloud. Even under the poor condition of network, data produced by sensors can be saved in local storage, making it enough time for user to settle the network problem with no affecting on the environment monitoring.