Monitoring networks optimization with simulated annealing

Nunes, Luís

http://hdl.handle.net/10400.1/1160

Use this identifier to reference this record.

Name:	Description:	Size:	Format:
TES2003a.pdf		2.27 MB	Adobe PDF	Download

Send Feedback

Authors

Nunes, Luís

Abstract(s)

In this work some methods to optimize environmental monitoring networks are proposed. These methods share simulated annealing as the approximation algorithm. Only monitoring networks reduction is treated here. Monitoring network optimization is a very actual problem given the large number of existing networks in many countries operating large numbers of stations, some of which may be redundant, with very high exploitation costs. Difficulties appear when exploitation costs pushes the dimension of a network towards a minimum, and the statistical reliability pushes in the opposite direction. Finding the optimal dimension may be a very difficult optimization problem due to the large number of combinations, even for small network dimensions. Further complications appear when the available data is too incomplete or come from different homogeneous areas. Some practical answers to these problems were sought in this work. Results showed that optimizing a monitoring network dimension and location of stations, without compromising the quality of the collected data, could attain large reductions in exploitation costs. Simulated annealing showed to be a very flexible and efficient algorithm.