Chemical sensor networks for the aquatic environment.

A recent commentary speaks to a broad range of potential applications for autonomous chemical sensor networks, including systems that monitor the state of the environment in real time.1 However, the authors of the commentary noted that existing sensor networks are “almost entirely restricted to transducers for detecting physical parameters such as temperature, pressure, light or movement”. They termed this the “chemical sensor paradox”, a condition that results because current technology “makes the realization of small, autonomous, reliable, chemo/bio-sensing devices impractical at present”. Rapid strides have been made toward autonomous chemical sensing capabilities in the past decade by the marine and aquatic chemistry communities. While chemical sensors are not at the same level of cost or reliability as physical sensors, a variety of chemical sensing systems are now continuously deployed in aquatic environments such as rivers, lakes, estuaries, and the open ocean. These chemical sensors are being operated, in some cases, for multiyear periods and in the ocean at thousands of kilometers distance from shore. Data from dozens of sensors are being delivered in nearreal time directly to the Internet. The primary emphasis of this review is focused on such chemical sensor networks that can be deployed on autonomous platforms in aquatic environments and then operated without significant human intervention for extended periods. The aquatic sensor networks that we consider have two essential characteristics: (1) chemical sensors are deployed in situ at multiple locations, and (2) the networks are operated for months to years. These observing systems are dedicated to observing environmental processes in real time. Much of this effort is quite recent, and this article serves as an introduction to this work, as well as a review. We focus on observing systems that operate for extended periods for two primary reasons. Long-term observations of the environment are essential to understanding variability driven by natural and anthropogenic climate change. In situ sensors also provide the continuous data needed to characterize high-frequency signals that cannot be easily sampled by manual methods. Undersampling of time-varying environmental processes can result in aliasing of the observed signal. The variability that is driven by processes occurring at higher frequencies than the sampling frequency of the environment can appear as unrecognizable low-energy events.2 Environmental chemists have often avoided the questions related to undersampling time-varying signals by assuming that aquatic systems are in a steady state. With this assumption, a single set of observations then provides an adequate assessment of environmental processes. This assumption has been necessary because, until recently, nearly all observations of chemicals in the aquatic environment required that a sample be collected and returned to the laboratory, where a variety of sophisticated tools could be used for sample analysis.3 Much of the ocean is sampled only a few times in a decade because of the long (weeks in some cases) transit times from seaports to mid-ocean regions. Even in the coastal ocean, or lakes and rivers, samples for chemical analysis are generally collected only at monthly intervals, if at all.4 Such sampling rates are often inadequate to characterize the dominant seasonal, daily, or semi-diurnal processes, which occur at well-defined frequencies, as well as episodic events driven by storms or other processes. Even the signals of decadal-scale processes may be contaminated † Monterey Bay Aquarium Research Institute. ‡ University of Washington. § North Carolina State University. 623 Chem. Rev. 2007, 107, 623−640