|
|
Automated monitoring of aquatic life in remote streams and rivers: Technical report project "Monitoring of aquatic life using a capacitive sensor array technique" Authors: T. Bach and B. Truyen Publication Date: Jun. 2007
Abstract: The monitoring of threatened and endangered aquatic species in remote streams and rivers requires frequent and accurate surveys at a variety of locations and times, in order to protect and enhance the viability of tenuous populations. Developments such as culverts and bridges require monitoring both before and after installation to detect adverse habitat impacts. Budgetary constraints require frugality in both development and monitoring projects. Present methods of aquatic life monitoring require large manpower investments and equipment expenditures, yet offer only modest data quality. May 2005, the United States Department of Angriculture, Forest Service, launched a call for the development and testing of 'equipment that can be used to directly detect, measure the size of, and/or differentiate the species of fish and/or other aquatic species, in their natural habitats, without manual intervention.' The ETRO-IRIS research group joint forces with Sensatech Research Ltd, to participate in this call, and won a $240.000 USD research contract to develop and test a new system for aquatic life monitoring based upon the principles of Electrical Capacitance Tomograohy. Electrical Capacitance Tomography (ECT) seeks to reconstruct internal distribution of electrical impedance contrast from the measurement of the boundary potentials arising from passing an AC current through an object. Sensatech Research Ltd, is a high-tech SME, located in Brighton, UK, that develops, produces and supplies custom electronic non-contact sensing solutions, primarily based on capacitive and electric field sensing techniques. Building upon its long standing tradition in the development of capacitive sensing technologies, Sensatech Research Ltd will be involved in the design of a new 2D capacitive sensing array. ETRO-IRIS will contribute its most recent research results on the derivation of a new class of subspace based structured problem formulations for the solution of ill-posed nonlinear inverse problems such as that of ECT.
The system developed in this project will basically consist of a flat 2-D array of electric field sensors in a "checkerboard" configuration, placed on the bottom of the streambed, and with the associated electronics either submerged or remote connecting to the array with a shielded ribbon cable. This system measures the complex impedance above each individual sensor pad and between any pair of pads, and uses appropriate software to determine the number, size and shape of fish. From this data, the desired specie information and the necessary noise discrimination from debris is extracted. Academically, this technology is referred to as Electrical Capacitance Tomography. This system has the advantage of being relatively inexpensive and mechanically robust, with the actual sensors consisting of insulated copper pads laid out on a printed circuit board, causing minimal effect on the stream flow. The sensors themselves can be considered a disposable item, and left in place between successive measurements, with the monitoring electronics only attached when data acquisition is required.
Electrical Capacitance Tomography, which is concerned with reconstructing a spatially varying conductivity distribution inside a bounded domain from partial knowledge of the Neumann-to-Dirichlet or Dirichlet-to-Neumann map, is known to be a notoriously difficult inverse problem, due its nonlinear and severely ill-posed nature. Despite the theoretical limitations and often disappointing performance, output least squares (OLS) methods continue to play a prominent role in most practical applications of ECT. Recently, ETRO-IRIS suggested a new variational solution method, which, unlike OLS, is guaranteed to deliver solutions that satisfy both the Thompson and Dirichlet variational principles associated with the ECT problem, irrespective of any additional smoothness assumptions on the conductivity distribution. Whereas conventional output-least squares methods can be regarded as minimizing a certain error norm, solutions are recovered here as the minimizers of a closely related residual norm problem. An iterative solution scheme is shown to lead to a sequence of structured sparse matrix problems, the conditioning of which appears to be far more favorable than typically observed in output least squares.
|
|
