This experimental activity focuses on configuring, maintaining, and adaptive power management of remote sensors and their embedded software used in environmental observing systems. The approach is based on a well known and popular cluster configuration system out of UCSD called Rocks. Sensor platforms and sensor networks require scalable and reliable solutions to manage their software configurations. This activity sets out to prove or disprove that a solution proven to work for heterogeneous cluster and data center configuration and management (ROCKS)can be applied to distributed sensors. Attached to this work is a set of integrated research and development activities on improved power management for sensor platforms and their operational workloads. Broader impact derives from the initial targeted communities in limnology and marine science with associated real world scenarios, and extends to other environmental observing systems as well as disaster response. Educational content from the work will be developed and inserted into a graduate level course at UCSD on embedded computing. Intellectual merit is found in the new approach to sensor reproducibility and adaptive power management algorithms.