DIMENSIONALITY PROBLEMS IN WATER-QUALITY NETWORK DESIGN

Dimensionality problems in design of water quality networks arise from the multiple variables measured, which may be generated by different physical, chemical, and biological processes, and from the multiple spatial and temporal dimensions of the processes generating the measurements. The problem is further complicated by the range of objectives which a network may be required to meet. The designer is almost inevitably constrained by budget considerations to employ the same network to measure an array of variables, often for more than one objective. These considerations serve to pose serious difficulties in arriving at an analytical formulation of the problem, far less to solve it, with the result that almost all existing water quality networks have been designed on an ad hoc basis. Although no one approach is likely to be universally applicable, several techniques have been employed in particular situations. Included are state estimation, time series analysis, probability theory, and the theory of regionalized variables. Several promising applications of these techniques to water quality network design are reviewed, and suggestions for future research directions are advanced. Copyright 1979 by the American Geophysical Union.