Model-Driven Approach to Optimization of Monitoring Designs for Multiple Water Quality Parameters

The issues of possible improvements, increased efficiency and/or optimization of monitoring systems in general, and monitoring designs, in particular, attract the attention of researchers for years. Application of formal techniques for these purposes looks appealing since it may validate suggested procedures or justify expenses required for data collection. The paper describes an approach to the development of sampling programs as a solution of the operation research model which minimizes the total number of water samples collected over an investigated period of time under the condition that the uncertainty of estimates derived from the monitoring data is kept below an acceptable level. Since concentrations of water constituents exhibit different variability, the numbers of observations required to achieve the same uncertainty level in their estimates vary significantly. In order to make a practically meaningful recommendation on the frequencies of observations, it is necessary to compromise temporal monitoring designs for all water quality parameters whose concentrations are derived from the same grab water sample. Given that concentrations of these parameters are formed under common hydrological and climatic conditions, it is reasonable to assume that series of concentrations are somehow related. It had been shown that if such dependencies are detected, they can be used to develop temporal monitoring designs common for water quality parameters determined from the same water sample and can significantly reduce the total number of observations required for water quality assessment. The proposed approach has been tested on observation data collected on a section of a small river in a highly urbanized area. The proposed approach may help to develop efficient monitoring designs with the reasonable cost of sampling programs by considering subsets of the water quality parameters.