Field investigations in the hydrologic sciences often are limited by the ability to collect data at the high spatiotemporal resolution necessary to build accurate predictive models or to control complex engineered systems in real time. Here, we describe LakeNet, an embedded wireless sensor network constructed by an interdisciplinary team of hydrogeologists, environmental engineers, and electrical engineers at the University of Notre Dame. Off-the-shelf temperature, dissolved oxygen, and pH probes are suspended from floating, waterproof cases with electronics, forming sensor pods. Wireless transmission to relay stations and a PC gateway enable researchers to interact with the network remotely to alter sampling patterns, download data, and analyze data trends using the gateway's recursive processing of raw data. LakeNet functions as a "smart" network, in which each pod is aware of surrounding pods. Ongoing research will allow in-network computation to detect change points in the data stream, thus triggering altered sampling strategy in response to sensed events or individual probe/pod failure. This ability for the sensor network to alter sampling strategy can increase sensor shelf life, which will become more important as novel sensors are developed, and decrease system energy requirements. Initial deployment at St. Mary's Lake on the Notre Dame campus in the fall of 2005 showed that temperature, pH, and dissolved oxygen changes were consistent with diurnal fluctuations in sunlight, hence, photosynthesis. In describing LakeNet, we summarize current capabilities and ongoing research, including some of the challenges that need to be overcome for further sensor network development, widespread deployment, and maximum usefulness.