Detecting acid precipitation impacts on lake water quality

The statistical objectives of the U.S. EPA programme for monitoring acid deposition in lakes included detecting temporal trends in water quality variables for single lakes and detecting trends in average conditions for groups of lakes. Statistical methods for trend analysis are considered. Historical data was used to establish the important characteristics of lake quality time series. Seven tests for trend were evaluated using Monte Carlo analysis. The best performing tests were the seasonal Kendall, modified t test and analysis of covariance on ranks. The power of trend detection and the effect of sample timing are considered. Common trends could be detected more quickly in groups of lakes than in single lakes.