Harvard Forest Research

Cyberinfrastructure for an ecohydrological sensor network

Principal Investigator: Emery Boose
Harvard Forest: Jan 01 2009 - Jan 01 2010:

Abstract:
This interdisciplinary project will develop cyberinfrastructure tools to support real-time studies of precipitation and stream flow at the Harvard Forest. The project will utilize an existing array of long-term meteorological and hydrological stations. Work will focus on (1) developing and deploying a mesh-based wireless network to provide connectivity to field sites, and (2) developing and deploying Analytic Web technology to support real-time data integration, quality control, analysis, and modeling.

(1) Sensor Network. Stream gages designed to measure surface flow (and to support other physical and chemical measurements) are critical to improving our understanding of the ecology and hydrology of small forested watersheds. The goal for this part of the project will be to deploy a wireless network to enable real-time observations from an array of stream and wetland gages currently in use in Harvard Forest. Low-cost, commodity wireless networks such as 802.11 and XTend as well as commodity embedded Linux controllers will be used to interface with the stream gages. Since little or no infrastructure is available at these locations, the network will need to subsist on battery (or solar) power and will need to employ aggressive energy optimization techniques to ensure long lifetimes. The wireless network will also need to employ mesh networking, possibly with long-range 802.11 links for the backhaul, in order to connect to the Harvard Forest campus and the Internet. The project involves a combination of field and laboratory work.

(2) Analytic Web. The concept of an Analytic Web was proposed nearly ten years ago by UMass computer scientists and Harvard Forest ecologists to address the growing problem of complex ecological datasets whose origins and production processes are insufficiently documented. The Analytic Web provides automated support for designing and executing the processes used to build complex datasets so that scientific results can be reproduced by other scientists. A prototype system (SciWalker) has been developed, with support from the National Science Foundation, and tested by application to eddy covariance studies of carbon flux at the Harvard Forest. The goals of this part of the project will be to continue development of SciWalker by improving support for process execution and auditing and by developing a user interface friendly to ecologists; and to use the enhanced SciWalker to build a real-time system for integrating, analyzing, and modeling meteorological and hydrological data at the Harvard Forest.
: