Skip to main content
eScholarship
Open Access Publications from the University of California

IDEA: Iterative experiment Design for Environmental Applications

  • Author(s): Amarjeet Singh
  • Maxim Batalin
  • Michael Stealey
  • Victor Chen
  • Mark H Hansen
  • Thomas C Harmon
  • Gaurav S. Sukhatme
  • William J. Kaiser
  • et al.
Abstract

This paper reports the first application of actuated sensing systems for high spatiotemporal resolution characterization of the threedimensional environment of river and lake aquatic systems. The development of a new method and its verification in these two application areas is described. Both applications involve dynamic phenomena - one resulting from flow of the water and the other from rapidly evolving biological processes. These applications are typical environmental monitoring problems. They exemplify the key challenge in such problems - characterizing phenomena displaying spatiotemporal heterogeneity. In many such examples, the application requires a diverse array of measurements based on sensors for physical, chemical and biological systems. Together, these requirements pose a significant challenge for conventional sensor network methods. We describe the development and applications of a new general purpose method for actuated sensing - Iterative experiment Design for Environmental Applications (IDEA). IDEA introduces a new in-field adaptation of the sensing systems including static and actuated sensors. IDEA addresses the limitations of previous sampling approaches, for example conventional adaptive sampling, by guiding adaptive sampling with an iteratively developed phenomenon model. This paper presents applications of IDEA to: (1) Three-dimensional characterization of contaminant concentration and flow at the confluence of two major rivers; and (2) Characterization of phytoplankton dynamics in a lake system. These applications provide ideal tests by presenting complex structures associated with each phenomenon and enabling a comprehensive evaluation of the general applicability of IDEA methodology. Improved performance using guided adaptive sampling is demonstrated for two existing methodologies, stratified adaptive sampling and hierarchical non-stationary Gaussian Processes. The IDEA experimental results both validate the general applicability of this method and also have advanced the understanding of interrelated physical, chemical and biological processes in these sampled environments.

Main Content
Current View