Use of extension variance in monitoring of fluoride in bottled water

Abstract

Abstract Temporal variograms allowed the analyzing of the temporal variance of eight sources of mineral waters during the four climatic seasons. The water sources are located in the state of São Paulo, Brazil. The extension variance compares the temporal variance obtained in the collection interval t with the temporal variance obtained in the collection interval T (where T is twice as large as t). Based on the calculation of the extension variance, relative sampling errors for the confidence intervals (CI) equal to 68% and 99% were obtained. For the sampled sources, the greater the sampling interval, the greater the values obtained for the extension variance and for the relative sampling error. The greater the confidence interval analyzed, the greater the relative sampling error to be obtained. The results showed a very high global sampling error for collection intervals greater than 32 days (relative error greater than 10%) when the confidence interval was CI = 68%. When the confidence interval was 99.9%, for collection intervals greater than two days, relative sampling errors greater than 10% have already been obtained. It was concluded that for the fluoride parameter the sampling time should not be longer than two days.