Effects of sampling frequency and laboratory detection limits on the determination of time series water quality trends

Time series trend analysis using the non‐parametric seasonal Kendall Sen Slope estimator and the seasonal Kendall trend test was conducted on a water‐quality data set (raw and flow adjusted) from part of the Wellington Regional Council's freshwater baseline water quality monitoring programme. The trend analysis was conducted a number of times to determine what effects changing the sampling frequency and detection limit may have on the overall result of the analysis. Results indicate that trends detected using a low detection limit are often not detected when a higher limit is adopted. This holds for upward and downward trends. Results also indicate that if the sampling frequency was changed from monthly to quarterly fewer trends were detected. The results also show that where the quarterly data shows a trend it is usually of a slightly different magnitude (slope) compared with the monthly data. This paper serves to illustrate the importance of considering detection limits of variables and sampling frequencies for state of the environment monitoring programmes.