TESTING FOR TRENDS WHEN THERE ARE CHANGES IN METHODS

Data from monitoring projects often include sampling or analytical cha nges that preclude trend analysis on the entire period of record. A mo dification of the nonparametric Kendall's test for monotonic trends, w hich accounts for such changes in the period of record, is described h ere. This approach blocks the data so that only data collected or anal yzed under similar circumstances are compared. Alternatively, when app ropriate data exist, data collected using the old method may be calibr ated to values expected from the new method. Traditional trend tests m ay then be applied to resulting data sets. Results from simulations as sessing both the power of the blocked test and the standard test perfo rmed on calibrated data are presented. The power of the blocked test e xceeded the power of the calibration approach only when the calibratio n error was extremely large. Both the blocking and calibration approac hes were applied to and compared for chemical data from Vermont lakes.