ANALYSIS OF SAMPLING FREQUENCY IN GROUNDWATER QUALITY MONITORING SYSTEMS - A CASE-STUDY

Sampling frequency is one of the most crucial factors in the design of groundwater quality monitoring systems. Monitoring systems in general have two major objectives: (1) to describe natural processes and long -term changes and (2) to serve as alarm-systems and detect single poll ution events. A comparison between two data sequences of different sam pling frequency - weekly and monthly - is made through an example of t he groundwater quality monitoring system in the karstic region of the Transdanubian Mountains in Hungary. Hydrogeochemical time series were first decomposed into their components: trend, periodicity, autocorrel ation, and rough in succession. in order to identify outliers within t he rough, Exploratory Data Analysis (EDA) was applied. Optimal samplin g frequency was determined based on the analysis of the above componen ts. Results have shown that: (1) seasons shorter than two months do ex ist in the studied time series which cannot be captured by monthly sam pling; (2) for monitoring seasonal processes samples should be collect ed at the Nyquist frequency (at least two samples per period); for pol lution detection autocorrelation lag-time (or semi-variogram range in time) should determine the sampling distance; in the lack of autocorre lation property the analysis of outliers should guide the sampling des ign; (3) cross-correlation analysis between precipitation and the obse rved parameters indicative of pollutant travel time yields valuable ad ditional information on the pollution sensitivity of the hydrogeologic al system.