Collecting field soil moisture information is the foundation of auto-irrigation. This paper introduced a soil moisture monitoring wireless sensor network (WSN) and with application background of automatic drip irrigation for cotton field. Firstly, application background was analyzed and application requirements were defined. The system worked together with a drip irrigation system in cotton field. In the second part, the frame of the system was given. It was composed of battery-powered field sensor nodes, solar powered WSN-GPRS gateway and data manage server. Hardware structure of the sensor node was introduced in details. Theoretical foundation of the design was introduced. After study, it was found that the soil moisture in a certain size of area was very nearly, which lead to the possibility of network data compression. In the third part, a novel data compression algorithm was proposed. In the algorithm, all the sensor nodes were encoded with Morton code and a logical multi-lays cluster was constructed among the nodes. Making use of the similarity of the output of the sensor nodes, the power cost was reduced significantly and prolongs life of the network. A power saving mode was introduced which made the network have a good flexibility. More sensor nodes had little impact on the network as Greedy Perimeter Stateless Routing (GPSR) algorithm was used to routing data. At last, some test and experiments results were shown in diagrammatic form. The system was field tested over six month period (April 2009 to Oct. 2009) in Xin-jiang province of China. The experimental results demonstrated that large scale, long-term collecting of soil moisture information can be achieved by using our proposed system.