Abstract: The data characteristics of forest environmental surveillance include a relatively slow changing rate over a long time period, high data redundancy and a sharp change following emergencies such as fire and rainfall. In partnership with the trials of wireless sensor network in forests, a long-term and effective system configuration was analyzed in terms of system architecture, sampling cycle of sensors and system hibernation based adaptive routing algorithm to capture changes in environmental variables in real time manner. Besides, network energy saving and system life-time prolonging were also examined. An adaptive sampling method with changing sampling cycles was proposed to address the issues of losing effective information due to long-term hibernation, and a self-adaptive dynamic routing by integrating system hibernation and clustering algorithm was also examined in the work. Results showed that compared with the conventional methods for monitoring forest resources and environmental parameters, the approaches presented here not only improve the depth of environmental data capture in temporal and spatial domains, but also elevate the speed of the system responds to changes in environmental variables. Apart from this, the proposed approaches and systems are energy-saving, and can maintain the operational stability and reliability, which are suited to forest environmental monitoring.