Efficient wireless sensor networks require inferential ecosystem models that can weigh the value of an observation against cost of transmission. Transmission costs make observations ‘expensive’; networks will typically be deployed in remote locations without access to infrastructure (e.g., power). Sampling intervals will range from meters and seconds to landscapes and years, depending on the process, the current states of the system, the uncertainty about those states, and the perceived potential for rapid change. The capacity to sample intensively makes sensor networks powerful, but high frequency data have value only at specific times and locations. Given that intensive sampling is sometimes critical, but more often wasteful, how do we develop tools to control the measurement and transmission process?
The measurement control process can involve optimization over the learning that can occur in multiple ecosystem models. In a given model, the value of an observation can be evaluated in terms of its contribution to estimates of state variables and important parameters. There will be more than one model applied to network data, including water, carbon, and energy balance, biogeochemistry, tree ecophysiology, and forest demographic processes. The value of an observation will differ among models. Inference is needed to weigh the contributions against transmission cost. Network control must be dynamic and driven by models capable of learning about both the environment and the network.
We discuss preliminary application of Bayesian inference to model data from a developing sensor network as basis for controlling the measurement process. Our example involves soil moisture, but we discuss plans for broader application of the approach.