Wireless sensor networks (WSNs) are widely applied in the supervisory control of agricultural fields, but their cost is an important limitation in their widespread application. This paper proposes a design and optimization scheme for monitoring plant cultivation involving deployment of multiple low cost sensing nodes combined with a few highprecision calibration nodes to reduce the construction cost of the system. All of the sensing nodes are configured using a low cost two-step data fusion (LCTDF) strategy to reduce the frequency of wireless data transmission. The amount of data transmission can be effectively reduced by optimizing the sensor nodes’ energy consumption, using the weighted data fusion mechanism to combine the homogeneous data of the sensing nodes. Each sensing node also uses an energy saving sleeping schedule. A testbed including the sensing nodes as well as a calibration node was also created and used for validating the basic functions of the proposed WSN system. Results of the demonstration experiment showed that the proposed LCTDF algorithm can be successfully implemented for environmental monitoring tasks. Additionally, the sleep scheduling of the sensor nodes also significantly improved the energy efficiency of the system.