Sensitivity of non-parametric trend analyses to multi-year extremes

Non-parametric tests for trend are widely used in analyzing hydrologic data because they presume nothing about the underlying distribution and are relatively insensitive to the presence of individual outliers. However, hydrologic data often include multi-year sequences of wetter or drier than normal periods. Such sequences may be episodic in nature and may persist for several years, but may imply no systematic change. While trend results are relatively insensitive to individual outliers, multi-year sequences of extremes can have a significant effect on the test results ifthose sequences occur near the beginning or the end of the period of record used for the test. The sensitivity of trend results to a sequence of extremes depends on the placement of that sequence in the period of record being tested and on the length of the sequence of extremes relative to the length of the record.