Evaluation of water quality monitoring networks using the concept of information entropy

Demonstrates the procedure and flexibility of using the information theory approach to evaluate the performance of water quality monitoring networks. Specific application of these are used in the evaluation of the Isaac River water quality monitoring network in Central Queensland.. Continued population growth, improved standards of living, and an increasing concern for the environment require careful management strategies to ensure the ecologically sustainable development and future availability of water resources. Water quality monitoring is an integral component of this management process. A water quality monitoring network is a monitoring system which provides management with the status of water quality conditions at a variety of locations and instances of time. While a range of significant challenges exist within the effective and efficient design of the monitoring system, one of the most critical of these challenges is the need for the selection of representative temporal and spatial sampling frequencies which neither replicate nor compromise the 'informativeness' and effectiveness of the monitoring. Recent research has employed information entropy concepts of transferred information and loss of information content to measure the representativeness and 'informativeness' of water quality monitoring networks. On the basis of a review of the principles and theory which underpin the concepts in their application to evaluation of performance, an improved methodology based on entropy theory was developed and applied to a typical Central Queensland water quality monitoring network. Success was particularly achieved in identifying the optimal selections of water quality variables and monitoring station locations. Other significant results from the analysis include the recognition of relationships between the transfer function and the criteria of transferred information, evidence of a 'priority order' for the removal of variables from a monitoring programme and a relationship between the co-variance matrix determinant and the class interval size which may alleviate the problem of calculating unstable joint entropies.